The University of Chicago the Response of Brucella Ovis

THE UNIVERSITY OF CHICAGO

THE RESPONSE OF BRUCELLA OVIS TO CHANGING AND CHALLENGING

ENVIRONMENTS

A DISSERTATION SUBMITTED TO

THE FACULTY OF THE DIVISION OF THE BIOLOGICAL SCIENCES

AND THE PRITZKER SCHOOL OF MEDICINE

IN CANDIDACY FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

COMMITTEE ON MICROBIOLOGY

LYDIA MIMI VARESIO

CHICAGO, ILLINOIS

JUNE 2021

To my dad, who taught me to love science and was always there for me, supporting me in my endeavors and choices, I miss you so much; and to Luca Daveggio, for everything you meant to me, for all that I owe you, for the amazing person you were. You will always be in my heart.

TABLE OF CONTENTS LIST OF FIGURES ...... vi LIST OF TABLES ...... viii ABSTRACT ...... ix ACKNOWLEDGEMENTS ...... xi 1. INTRODUCTION ...... 1 1.1. Preface ...... 2 1.2. Bacterial growth ...... 2 1.2.1. Stationary phase ...... 4 1.2.2. Stringent response ...... 5 1.2.3. The general stress response ...... 7 1.3. The a-proteobacteria class ...... 8 1.4. The pathogen Brucella ...... 9 1.5. Pathogenesis...... 11 1.6. The intracellular niche ...... 13 1.6.1. Nutrient limitation within the BCV ...... 15 1.6.2. The respiratory burst and oxidative stress ...... 16 1.6.3. Sulfur metabolism and pathogens ...... 20 1.7. Brucella ovis, the runt of the litter ...... 22

1.8. Brucella cultivation and CO2: a brief history ...... 24 1.8.1. Natural variation in the requirement for CO2 supplementation ...... 26

1.9. CO2 metabolism and carboxylases ...... 27 1.10. Carbonic anhydrases ...... 29 1.10.1. Biological function of CO2 and carbonic anhydrases ...... 31 1.11. Conclusions ...... 32 2. A CARBONIC ANHYDRASE PSEUDOGENE SENSITIZES SELECT BRUCELLA LINEAGES TO LOW CO2 TENSION ...... 34 2.1. Preface ...... 34 2.2. Introduction ...... 34 2.3. Results ...... 34 2.3.1. A forward genetic selection identifies B. ovis mutant strains capable of growth in an unsupplemented atmosphere ...... 34 2.3.2. A bcaA frameshift enables growth in an unsupplemented atmosphere ...... 36 2.3.3. bcaA frameshift mutations are the sole genetic mechanism by which B. ovis growth is spontaneously rescued under low CO2 tension ...... 36 2.3.4. The frameshifted alleles, bcaA1BOV-4BOV, are dominant over the wild-type bcaABOV allele ...... 37 iii

2.3.5. Expression of E. coli b-carbonic anhydrases is sufficient to enable B. ovis growth in a standard atmosphere ...... 38 2.3.6. A comparative analysis of bcaA sequences and the CO2 growth requirement in the genus Brucella ...... 39 2.3.7. Functional B. abortus bcaA alleles are rapidly selected from a bcaABAB pseudogene under CO2 limitation...... 40 2.3.8. CO2 limitation triggers a large-scale transcriptional response in wild-type B. ovis, while transcription in the bcaA1BOV strain is insensitive to CO2 shifts ...... 42 2.3.9. CO2 limitation induces a stringent-like starvation response and activates transcription of the virB type IV secretion gene cluster in wild-type B. ovis ...... 43 2.3.10. Pseudogene sequences, including bcaABOV, are highly conserved across all B. ovis isolates ...... 45 2.4. Discussion ...... 47 2.4.1. Brucella cultivation and the resurrection of a carbonic anhydrase pseudogene ..... 47 2.4.2. CO2 limitation elicits a starvation and virulence gene expression response ...... 50 2.5. Materials and methods ...... 51 2.6. Summary ...... 59 3. BRUCELLA OVIS CYSTEINE BIOSYNTHESIS CONTRIBUTES TO PEROXIDE STRESS SURVIVAL AND FITNESS IN THE INTRACELLULAR NICHE ...... 60 3.1. Preface ...... 60 3.2. Introduction ...... 60 3.3. Results ...... 61 3.3.1. B. ovis cysE Tn-himar mutant strains have a fitness defect in stationary phase ..... 61 3.3.2. B. ovis ∆cysE enters stationary phase prematurely and has reduced culture yield in vitro ...... 62 3.3.3. cysK1 and cysK2 function redundantly in cysteine biosynthesis ...... 63 3.3.4. B. ovis ∆cysE and ∆cysK1 ∆cysK2 strains are sensitive to exogenous H2O2 stress 64 3.3.5. ∆cysE and ∆cysK1 ∆cysK2 have reduced viability in the intracellular niche ...... 64 3.4. Discussion ...... 66 3.4.1. A genome-scale search for B. ovis stationary phase mutants leads to cysteine metabolism ...... 66 3.4.2. Cysteine, glutathione and hydrogen peroxide stress ...... 68 3.4.3. Cysteine and growth and the intracellular niche ...... 69 3.5. Materials and methods ...... 70 3.6. Summary ...... 76 4. DISCUSSION, CHALLENGES FOR THE FUTURE, AND CONCLUSIONS ...... 77

4.1. CO2 sensitivity ...... 77 4.2. Other Brucella carbonic anhydrases ...... 79 4.3. Effects of altered sulfur and cysteine metabolism in B. ovis...... 82 4.3.1. Coloration of ∆cysE strains ...... 82 4.3.2. Sensitivity to transition metals ...... 84

4.4. Cysteine and methionine metabolism in Brucella ovis ...... 84 4.4.1. B. ovis presumably does not encode for genes required for the reverse transsulfurylation pathway ...... 84 4.4.2. The redundancy of the two cysK genes in B. ovis ...... 85 4.5. Conclusions ...... 86 5. APPENDIX A – REGULATION OF THE ERYTHROBACTER LITORALIS DSM 8509 GENERAL STRESS RESPONSE BY VISIBLE LIGHT ...... 88 5.1. Preface ...... 88 5.2. Introduction ...... 88 5.2.1. LOV-HWE kinases: An overview ...... 88 5.2.2. Erythrobacter litoralis DSM 8509 as a system to study LOV-HWE kinase signaling ...... 90 5.3. Results ...... 91 5.3.1. A brief comparison of Erythrobacter litoralis DSM 8509 to other Erythrobacter species ...... 91 5.3.2. Whole-genome sequencing and development of genetic tools in DSM 8509 ...... 92 5.3.3. E. litoralis DSM 8509 LovK is a photosensor ...... 93 5.3.4. The E. litoralis DSM 8509 GSR regulon ...... 93 5.3.5. Light-dependent regulation of transcription at the genome scale ...... 95 5.3.6. Regulators of the GSR signaling pathway ...... 96 5.3.7. The role of three HWE-family sensor kinases in regulation of GSR transcription. 99 5.4. Discussion ...... 101 5.4.1. Light, LOV, and bacterial stress responses ...... 102 5.4.2. LovK functions as part of a consortium of GSR sensor histidine kinases ...... 103 5.5. Materials and Methods...... 105 5.6. Summary ...... 115 6. APPENDIX B – FIGURES ...... 116 7. APPENDIX C – TABLES ...... 158 8. REFERENCES ...... 167

LIST OF FIGURES

Figure 6.1 - Schematic representation of a growth curve ...... 116

Figure 6.2 - Growth of B. ovis harboring bcaA1BOV-4BOV alleles in an unsupplemented atmosphere ...... 117

Figure 6.3 - A single nucleotide deletion at the 3’ end of bcaABOV enables B. ovis growth without CO2 supplementation ...... 118

Figure 6.4 - Comparison of B. suis BcaA and E. coli b-carbonic anhydrases to BcaABOV and BcaA1BOV ...... 120 Figure 6.5 - Heterologous expression of two different Escherichia coli b-carbonic anhydrases enables growth of wild-type B. ovis ATCC 25840 without CO2 supplementation ...... 121 Figure 6.6 - Sequence polymorphisms in BcaA orthologs across the genus Brucella ...... 122 Figure 6.7 - Analysis of sequence and function of B. abortus bcaA alleles in B. ovis ATCC 25840 ...... 123 Figure 6.8 - RNA-seq experimental set up and measured gene expression changes in B. ovis bcaA1BOV upon CO2 downshift ...... 124

Figure 6.9 - Gene expression changes in wild-type B. ovis and B. ovis bcaA1BOV upon CO2 downshift from 5% to 0.04%...... 126 Figure 6.10 - KEGG pathway assignment of genes regulated in wild-type B. ovis in response to CO2 downshift ...... 127

Figure 6.11 - Heat map representation of a subset of genes regulated by CO2 downshift in wild-type B. ovis ...... 128 Figure 6.12 - Nucleotide differences between sets of functional genes and pseudogenes in B. ovis ATCC 25840 and their B. abortus ATCC 2308 orthologs normalized by gene length...... 129 Figure 6.13 - Fitness profile of B. ovis transposon insertion mutants as a function of growth phase...... 130 Figure 6.14 - Assessment of B. ovis mutant strain fitness as a function of growth phase identifies cysE as a determinant of stationary phase fitness ...... 131 Figure 6.15 - Functional classification of B. ovis genes whose disruption significantly impacts fitness during growth in Brucella broth ...... 132 Figure 6.16 - ∆cysE enters stationary phase prematurely; this growth defect is rescued by addition of cysteine to the growth medium ...... 133 Figure 6.17 - The stationary phase phenotype of a ∆cysK1 ∆cysK2 double deletion phenocopies ∆cysE and is rescued by cysteine ...... 134

Figure 6.18 - B. ovis ∆cysE is sensitive to H2O2 treatment; ∆cysE growth defect and peroxide sensitivity is mitigated by glutathione ...... 135

Figure 6.19 - B. ovis ∆cysE has reduced fitness in the intracellular niche of human macrophage-like cells and an ovine testis epithelial cell line ...... 136 Figure 6.20 - Recovered CFUs decrease at 72 hrs post infection...... 137 Figure 6.21 - Genetic complementation of THP-1 and OA3.ts infection ...... 137 Figure 6.22 - B. ovis ∆cysE is more attenuated in THP-1 human macrophage-like cells than in ovine testis OA3.ts cells ...... 138 Figure 6.23 - B. ovis ∆cysE strain is attenuated in Raw 264.7 macrophages ...... 138 Figure 6.24 - Brucella ovis and B. ovis bcaA1 have no difference in morphology ...... 139 Figure 6.25 - Modelling of BcaA structure in B. ovis and comparisons to experimental structures of other carbonic anhydrases ...... 139

Figure 6.26 - BcaABOV does not purify in a soluble form and is mainly found in inclusion bodies ...... 140

Figure 6.27 - BcaB from B. abortus cannot rescue B. ovis growth in 0.04% CO2 ...... 141 Figure 6.28 - Deletion of cysE leads to differences in coloring and tolerance to different metals ...... 142 Figure 6.29 - Supplementation of methionine does not restore ∆cysE growth to wild type levels ...... 143 Figure 6.30 - Predicted cysteine metabolism in B. ovis ...... 144 Figure 6.31 - Similarities of conserved regions of OCBS and OASS to B. ovis CysK1/2, and effect of cystathionine supplementation...... 145 Figure 6.32 - Core regulators of the Alphaproteobacterial general stress response ...... 145 Figure 6.33 - GSR regulators and their expression in E. litoralis DSM 8509 ...... 146 Figure 6.34 - Absorption spectroscopy provides evidence that LovK protein can function as a photosensor ...... 147 Figure 6.35 - General stress response regulon of E. litoralis DSM 8509 ...... 148 Figure 6.36 - Relative expression of genes involved in phototropy ...... 150 Figure 6.37 - The E. litoralis light-dark regulon overlaps with the GSR regulon...... 151 Figure 6.38 - Palindromic sEcfG binding site lies between nepR-ecfG and gsrP ...... 152 Figure 6.39 - Gene structure, protein domain structure and RNA-seq expression values for two additional unnamed HWE kinases encoded in the DSM 8509 genome ...... 153 Figure 6.40 - Target and control genes used for qRT-PCR analysis of GSR transcription ...... 154 Figure 6.41 - Combinatorial control of GSR transcription by three HWE-family sensor histidine kinases ...... 155 Figure 6.42 - Complementation of GSR transcription defects in strains with single deletions of GSR regulators ...... 156 Figure 6.43 - Model of GSR regulation in E. litoralis DSM 8509 ...... 157

vii

LIST OF TABLES

Table 7.1 - Growth of independent spontaneous mutants derived from forward genetic a selection grown in an air incubator (0.04% CO2) after isolation ...... 158 Table 7.2 - Sequence polymorphismsa between 16 independent B. ovis mutants that grow b without CO2 supplementation and the wild-type parent strain ...... 159 Table 7.3 - B. ovis ATCC 25840 Tn-Himar library ...... 161 Table 7.4 - B. abortus biovar association with bcaA allele clusters ...... 162 Table 7.5 - Gene and pseudogene comparison across B. ovis and B. abortus ATCC 2308 strains ...... 163 Table 7.6 ...... 165 Table 7.7 - Genome characteristics of select Erythrobacter spp. isolates ...... 166

viii

ABSTRACT

Bacteria inhabit many environments, from hot dry desert regions to high pressure niches on the ocean floors, from high salt areas to cold latitudes, from inside plant root nodules to inside human cells. Studying the kinds of habitats different bacteria encounter and how they react to the challenges these environments offer is important to understand bacteria in their natural state, what kind of responses characterize them and how they can function when confronted with adversity.

Here I present my research on Brucella ovis, a facultative intracellular pathogen that has evolved to withstand stressful and harmful situations, from nutrient limitation, to oxidative stress, to drastic drops in pH. One ambient factor that perturbs B. ovis homeostasis is the atmospheric level of carbon dioxide, as B. ovis cannot be cultured in laboratory conditions without CO2 supplementation. I examined the genetic underpinnings of the CO2 dependence of B. ovis growth, identifying mutations in a carbonic anhydrase gene (bcaA) as responsible for this particular metabolic requirement. B. ovis harbors a unique, non-functional pseudogene allele of bcaA

(bcaAbov), and I found that some B. abortus lineages also harbor a bcaA pseudogene, thus rendering them dependent on CO2 supplementation for growth. My data explain why B. ovis and select strains of B. abortus require elevated CO2 levels for growth, which was first noted over a century ago. Transcription of one third of the genes in wild-type Brucella ovis change when cells are shifted from high to low CO2 conditions; gene expression is unchanged upon CO2 shift in a strain in which the pseudogene is restored to a functional bcaA. Thus, wild-type B. ovis has increased sensitivity to environmental levels of CO2 because of a carbonic anhydrase pseudogene. This sensitivity could help B. ovis better detect when it is inside and outside the host.

I also present work in which I analyze Brucella ovis in the context of stationary phase, as a proxy to better understand the environment and related response that this pathogen encounters

ix within the host. I discovered that cysE -which encodes for a serine O-acetyltransferase, involved in the first step of de novo cysteine biosynthesis- is required for B. ovis fitness in stationary phase.

Deletion of cysE increases sensitivity to hydrogen peroxide and attenuates Brucella ovis in tissue culture infection models. Thus, sulfur and cysteine metabolism is important for resistance to the hostile environment within the intracellular niche and presents an intriguing target for drug development against Brucella infections.

ACKNOWLEDGEMENTS

This dissertation is the celebration and culmination of more than five years of work. There are many people I would like to acknowledge as they have been pivotal to my success, provided much needed support, or have helped out in specific moments on particular issues where I needed it the most.

First of all, I would like to thank my PI and mentor, Sean Crosson. He has guided me for the past five years, teaching, coaching, steering and supporting me along the way. I have learned so much from this monumental person, and am exceedingly grateful that I got the chance to work, discover, and thrive in his lab. Being Sean’s graduate student has made me a better scientist and gifted me the opportunity to freely explore my scientific interests, ask questions, discuss data and models, approach novel techniques and rediscover historical perspectives, and overall work in a positive, curious, and helpful environment he creates. Thank you mentoring me and all the time and patience you dedicated to my growth and learning.

I cannot think of the Crosson lab without Aretha. Aretha Fiebig is an amazing person and scientist. I’d like to thank her for the invaluable help and all the things she taught me throughout the years, for the long ‘quick questions’ that often derailed in topics more or less pertinent to the original query, for her understanding nature and endless patience, for her creative solutions and willingness to listen, for her openness in providing feedback and advice. She is one of my role models, both as a person and as a researcher, and has been fundamental to my betterment.

Julien Herrou was my scientific guardian angel for the longest time. He walked me through experiments and techniques, providing feedback and scientific discussion, mentoring me and dissuading me from falling into endless rabbit holes. He has been an amazing colleague and friend,

xi helping me stay focused and on track especially through the inevitable rough patches during my graduate career.

I would like to also acknowledge my former post doc mentor, Jonathan Willett, with whom

I worked during my rotation in the Crosson lab and whose research in Brucella engrossed me, eventually leading to my joining the lab and picking up where he left off. His approachable manner and good nature eased me into the microbiology lab, which was an unfamiliar and novel environment for me at the time.

I would like to deeply thank David for the intense scientific discussions and teachings, for often challenging thoughts and results, therefore helping in the development of models, theories and data interpretation in more concrete, solid, and reasonable form, broadening my understanding and interpretation of data. And for being my honest friend, always happy to drink a beer and chat whenever I needed to. I would also like to thank Ben Stein, who moved with Sean, Aretha, and myself. Especially in East Lansing, Ben has helped me out both inside the lab and out, always lending an ear to let me vent and providing support and counsel, as well as investing time and energy in helping me out with scientific problems as well as day to day hassles. In East Lansing, I also found a friend and exceptional colleague in Patrick McLaughlin who has an amazing personality and whose strong will has fed well into mine, uplifting my mood and challenging my ideas in unique ways.

I would also like to generally acknowledge the Crosson lab, past and present members. It has morphed greatly during these five years, especially with the move, but there has always been inner lab support, fruitful discussions and advice during lab meetings as well as at the bench. From

Chicago, I would like to specifically mention Daniel Eaton, for the Saturday lab laughs; Leila

Ruiz-Reyes, for all the soccer games and comradery, and Maggie Zhang, for her shining words of

xii encouragement and pure niceness. I also want to thank the newer lab members that have helped ease the move and create a newly positive, rich, collaborative, and supportive environment in East

Lansing, especially Hunter North, for her openness, positive energy, and for being an overall great person; Sergio Hernandez-Ortiz, for sharing music, games, and delicacies with me; and Tom Kim, for his serene presence and helpful manner.

I would like to deeply thank my Thesis Committee, for all the help they have provided in walking me through my graduate career, the scientific insights and advice, the encouragement and forward nudges: Glenn Randall, who believed in me from the start, and always has had time for me, with a smile; Maureen Coleman, who has always spontaneously offered me help and a place in her lab if needed; Lucia Rothman-Denes, whose honesty and affectionate manner have given me courage in the face of hard choices and solace in feeling supported and understood. I would also like to recognize my former Committee member, Howard Shuman, from whom I have learned so much, and whose wisdom and guidance were deeply appreciated.

Furthermore, I would like to acknowledge Matt Zurensky, Eric McLean, and Daniel Czyz, who helped me attempt a series of tissue culture experiments during my stay at the University of

Chicago. I would also like to thank Gabriel Vargas, who has been at my side both scientifically, providing feedback, advice, and a constant willingness to chat about science, as well as for being an awesome, supporting, encouraging, and unwavering true friend.

I cannot imagine my time in Chicago without mentioning all the people that have been walking through this same career path with me, for its entirety or part of it, and have helped alleviate the harder moments by sharing games, sports, trips or just simple evenings and dinners, making this whole experience complete, joyous, and worthwhile: Shan Kasal, Chris Stamper,

Andrew Miller, Audrey Williams, Alex Hoffman, Liana Hernandez, Zach Early, Ryan Duncombe,

xiii

Roy Morgan, Steven Erickson, Will Reidl, Matt Reyer, and Liz Ziegler. I would also like to mention a handful of friends that were not directly part my graduate cohort, that I met during orientation, that have unfailingly played pick up soccer with me, helping me vent my frustrations as well as providing cheer and merriment over nice cold beers: Marc Gillard, Paolo Andrich, Cem

Randa, and George Adams. I would also mention Liz Lee and Alan Linton, for awesome broomball games and impressive comradery on those bitter cold nights.

There are a number of people that are not part of the University of Chicago that have nevertheless contributed to my success and happiness while pursuing my graduate degree, as well as helping me in the many years that lead up to the start of my PhD. The first person I want to thank, and that I can never thank enough, is my dad. There are no words to describe how much I owe him. He has taught me to love and appreciate nature, at all levels, as well as research, to ask questions, to be curious and search for answers. He has always prompted me to argue my points, discuss my ideas and opinions, to not shy away from learning new things and to hold my head high. He always wanted me to believe in myself, to trust in my instincts and knowledge, to be independent and strong. He has given me so much encouragement and support while also letting me walk on my own, make my mistakes, and grow from them.

I would also like to thank my group of lifelong ‘genovese’ friends that I left behind in Italy when I moved to the US. They have always been there for me. For long phone calls dealing with whatever problem in the universe we feel like discussing, to high end chats about politics, philosophy, sociology, art, and psychology to basic arguments about sex, drama, and gossip, to remembering trips, summers, and weekend outings, to talking about everyday problems, sad moments, frustrations, and angst, helping each other out how best we can and always, always ready to lend a hand, an ear, time. There are so many of you I would like to acknowledge, as you have

xiv unwaveringly encouraged, supported and sustained over these years. Marc Bosi, Emiliano

Montermini, Salvatore Canto, Giorgia Manavella, Davide Porzio, Martina Peloso, Francesco

Parodi, Alessandro Gabbiani, Federica Camoirano, and Luca Villa, to name a few. Not from

Genova, but still dear to my heart and just as important in these five long years, are those friends who are scattered throughout Piedmont, and who have given me amazing support and affection:

Marco Ferrero, Camilla Fontana, Riccardo Pelle, Mauro Adriano, Sonia Palumbo, Danilo

Monateri, and Jessica Dardano; as well as my colleagues and friends from my Master’s lab, who contributed to my sanity and formation that eventually lead me to the Crosson lab, and have been willing participants to my rants and jests since: Teresa Poggio and Matteo Menotti. A special mention to Manuela Sigolo, the strongest person I know, genuine and compassionate, that has always reached out to me and cared for me at a distance. Thank you for everything.

I would like to finally acknowledge my family, first of all my mother, for teaching me strength and endurance, persistence and precision. For loving me and missing me, always taking care of me and for letting me know that I always have a home with her. I would also thank my sister, my blonde tall sister, for she is an incredible person, has forgiven my pigheadedness and has given me support and understanding, more than I deserved. I would like to thank the rest of my family, aunts, uncles and cousins, for the unconditional love and support you each have shown me all these years, the warm embrace and genuine smiles. A special thank you to my aunt

Giovanna Eva, who is bitingly direct and loudly exuberant, and one of the closest confidants I have, who has aided me time and time again when I was in need, never flinching, never hesitating, always 100% there, listening, counseling, and uplifting. When I think of family, I cannot forget my new family, and I want to acknowledge the MacNabb family, especially Martha and Tom.

They have embraced me as a daughter, taken care of me through the loss of my dad, knee surgery,

xv an unexpected move, and the many smaller hardships that these years have brought. They have celebrated my successes and have simply been two of the most generous, warm, affectionate people I ever had the fortune to meet. Thank you, so much.

Finally, I would like to thank Brendan MacNabb. Everything I thanked people up to now for, I also owe Brendan. He has been my anchor, my support in these wacky years, through the good and the bad. He has always stopped to listens without dismissing my concerns, has always tried to envision the best choices for me above his wishes, has given me all his support and held my hand through the many hardships. He has given me scientific advice. He has celebrated joyous days. He stood at my side during hard decisions. He took care of me when I needed it the most.

He has understood my intense and impetuous nature. He is an incredible friend, a knowing colleague, a cunning teammate, an adventurous travel mate, and, obviously, an irreplaceable partner.

xvi

1. INTRODUCTION

All over our planet, bacteria inhabit an immensely broad collection of environments. They colonize a more diverse set of habitats than any other organism on earth. They have adapted –and are adapting– to their niches in a constant arms race with challenges set to them by that very niche: whether it be nutrient limitation, predation, dehydration, pH changes, osmotic shock, high pressure, hostile host attacks, radiation damage, or antibiotics (to name a few), they constantly have to retain the capability to fight or cope with these conditions. Bacteria rise to such challenges by sensing the offending environments and consequently adapting their physiology. These changes can be regulated and temporary (transcriptional, translational or post-translational) or may be permanent (i.e. genetic mutations). The latter may be then fixed within a population if it results in an overall fitness advantage for that population. There is a balance between plasticity and rigidity in the capacity of bacteria to survive and thrive in their environments, and different bacteria have different tools that enable them to survive their current environment and react to eventual changes.

Thus, when cultivating bacteria in a laboratory, the conditions of growth and physiology examined often do not mirror the ‘natural’ living conditions encountered by that particular organism.

Studying the nature of the different environments that bacteria face and how they respond to these conditions is paramount to understanding how bacteria thrive and interact, and allows us to delve deeper into what characteristics are important for them to flourish in their natural habitats.

In this dissertation, I describe how different environmental factors, specifically partial CO2 pressures and sulfur-containing metabolites in the intracellular niche impact the fitness and possibly the evolution potential of the facultative intracellular bacterium, Brucella ovis. This

1 introductory chapter first presents key aspects of bacterial growth and some of the conserved responses that bacteria elicit upon different environmental perturbations. I next highlight characteristics of the Alphaproteobacteria, the class that Brucella belong to, before diving into details of Brucella physiology and infection biology. Since my thesis research involved investigation of host-derived stressors that impact Brucella, I will briefly cover a few details of the kinds of attacks Brucella face both within the host and specifically within the intracellular environment, and how this furtive microbe counteracts these challenging situations. As cysteine metabolism is one of the key features that enables Brucella ovis to survive host challenges, I will also detail aspects of cysteine and sulfur metabolism in my introduction. I will further describe

Brucella ovis and highlight physiologic features that set it apart from other Brucella species, outlining why these characteristics are important and warrant further exploration. Finally, I will describe the importance of carbon dioxide in Brucella metabolism, the enzymes involved in CO2 assimilation and their relationship to essential metabolic processes.

1.1. PREFACE

Some of the following introductory content has been adapted from two of my publications:

- Varesio LM, Willett JW, Fiebig A, Crosson S., Journal of Bacteriology (2019)

- Varesio LM, Fiebig A, Crosson S., Infection and Immunity (2021)

1.2. BACTERIAL GROWTH

Bacteria possess sophisticated systems to sense their environment, and equally sophisticated methods to respond to it. They are rarely in ideal, nutrient rich situations like those we fabricate in laboratory conditions. Measuring bacterial growth is a good proxy to assess if the conditions they are in are well tailored to the organism in question.

One of the most obvious struggles for bacteria is to survive starvation. Nutrient scarcity is a common daily challenge that bacteria face. Nutrient limitation may be caused by natural environmental paucity or by sequestration from more fit competitors. When grown axenically in batch culture (i.e. where the nutrients are added at the start of the experiment), the depletion of available nutrients shapes the growth curve, which is divided into several phases. Initially, the cells are in lag phase, whose duration varies based on the conditions the bacteria were in prior to commencement of the growth experiment as well as the species/strain in question. The lag phase is followed by the log phase. During this second phase, bacteria are doubling exponentially.

Physiologically, as they are actively growing and dividing, they favor energy-demanding reactions to synthesize nucleic acids, proteins, and other cell components. Once nutrients start to run out, or in response to other environmental changes that suppress growth (for instance a shift in pH), cells enter stationary phase (1). This phase is characterized by a gradual reduction of growth rate to zero and no net increase in cell numbers. Cells can persist in stationary phase for varying degrees of time in a more or less dynamic state, which depends on a multitude of factors, such as the starvation conditions (2). Eventually, cells pass from stationary phase to the fourth state, known as the death phase. Here, cells are rapidly dying, as can be assessed by colony forming unit (CFU) counts of viable cells, and up to 99% of cells in the population may perish. Whether this occurs stochastically or by programmed cell death (bacterial apoptosis) is not fully understood, but there is a small percent of the population that persists. These cells are now in what is known as the long-term

3 stationary phase, where a small population of viable bacteria can be detected. This is a dynamic state, where cells may die and consequently feed sister cells that can undergo some replication cycles or simply lay dormant without perishing. During this last phase, there is a particular signal phenotype associated with a series of morphological, physiological and transcriptional changes collectively known as the growth advantage in stationary phase (GASP) that allow cells to survive stress and persist in this final phase even for years (Figure 6.1) (3). The sensory pathways and responses needed for bacteria to survive in a persister state is poorly understood and currently an active area of study.

1.2.1. Stationary phase

Stationary phase is the period during growth in batch culture in which exponentially dividing cells slow growth and reach a plateau where there is no net increase in cell numbers.

Stationary phase was initially studied in the model organism Escherichia coli, and entry into stationary phase has often been studied in the context of nutrient starvation (1, 2). Upon entry into stationary phase, cells undergo morphological changes in cell shape and size (4), increased resistance to heat shock and hydrogen peroxide (H2O2) (5), high tolerance to elevated osmolarity

(6), release of nucleobases in the environment (7), decrease in protein synthesis, drastic metabolic shifts from biosynthetic metabolism to energy conservation, and overall global transcriptional, translational and post-translational modifications (2).

Studies in E. coli identified major central players of stationary phase induction. In 1990,

Lange and Hengge-Aronis discovered an alternative sigma factor, named sigma S (sS, for stationary phase, or starvation) encoded by rpoS (8). Transcription (8) and stabilization (9) of RpoS is induced upon entry into stationary phase (i.e. reduction of growth rate), which in turn affects transcription of an immense regulon – more than 1000 genes in E. coli (10–12). RpoS regulation

4 occurs at multiple steps (13). rpoS is regulated at the transcriptional level by growth rate (and in some form by ppGpp, see below) and cyclic adenosine monophosphate (cAMP), an important second messenger (14). Another major player in regulating entry into stationary phase is the hfq gene, which encodes for host factor protein 1 (HF-1 or Hfq) (15). Hfq is an RNA-binding protein that was shown to affect RpoS levels by binding to the sigma factor in E. coli (16), indicating that

Hfq is upstream of RpoS. It is now appreciated that Hfq is a global post-transcriptional regulator, affecting targets other than RpoS. Other molecules have been found to regulate transcription of rpoS in E. coli, such as H-NS (which inhibits Hfq and consequently blocks translation of rpoS mRNA) and OxyS (which also affects translation of rpoS by affecting the ribosome binding site)

(17). RpoS is not the only regulatory molecule that regulates transcription during stationary phase.

Depending on what commenced entry into this growth phase, other rpoS-independent proteins may come into play. For instance, other alternative sigma factors seem to play an important role, like rpoH (s32 or sH) or ecfG (SigT) (see below) (1).

Stationary phase is not only induced by nutrient limitation. Accumulation of toxic metabolites, osmotic shock (18), various stresses (such as oxidative stress, see below) may also drive entry into stationary phase, although nutrient starvation is the most extensively studied.

These different cell states are interconnected and partially overlap, based on the signals involved and on the bacterial species, allowing bacteria to react to different environmental cues and resist various stressful situations.

1.2.2. Stringent response

The stringent response is a conserved starvation response in bacteria. Different kinds of starvation, such as amino acid, fatty acid or, carbon depletion can trigger the stringent response, as well as osmotic stress (18–21), and they vary depending on the organism in question. The

5 immediate response is the rapid biosynthesis of the alarmone guanosine tetra- or penta-phosphate,

(p)ppGpp (22). In Escherichia coli, where the stringent response was first studied, RelA or SpoT are responsible for synthesizing (p)ppGpp in response to different kinds of nutrient limitation (lack of amino acids in the case of RelA; fatty acid, iron, and carbon starvation in the case of SpoT).

RelA binds RNA polymerase (RNAP) and, sensing the accumulation of uncharged transfer RNA

(tRNA) molecules in the A site of RNAP as consequence of amino acid starvation (23), synthesizes

(p)ppGpp. SpoT responds to different kinds of starvation, such as carbon, iron, and fatty acid limitation and also possesses a hydrolytic domain, so it is responsible not just for (p)ppGpp synthesis but also for its turnover (24). Bacteria have evolved varied mediators of the stringent response; in the Alphaproteobacterium Brucella (25, 26), for instance, a relA/spoT-homolog (rsh) replaces RelA and SpoT, and is the sole driver of (p)ppGpp synthesis and degradation.

(p)ppGpp was originally considered a master downregulator, as it negatively affects the half-life of most promoters thus repressing transcription, and in E. coli, increases levels of the alarmone to induce a decrease in stable RNA (tRNA ribosomal RNA (rRNA)) transcription as well as inhibition of DNA synthesis (27). It is now understood that (p)ppGpp is not only a negative regulator. Indeed, it also promotes transcription of rpoS (9) as well as of amino acid biosynthesis genes (28). Further studies have shown that in E. coli, more than 500 genes are differentially expressed in response to (p)ppGpp, the majority of which are upregulated (29). Of note, not all responses to the induction of the stringent response are the same in all bacteria. For instance,

Caulobacter crescentus, another a-proteobacterium, does not react to amino acid starvation by synthesizing the alarmone (30), but activates the stringent response when glucose (31) and ammonium (32) are limmiting. Although the stringent response has been mostly characterized within the Proteobacteria phylum, recent studies have also uncovered the effects of the stringent

6 response in other groups of bacteria. For instance, the Gram-positive Staphylococcus aureus encodes for the RelA/SpoT homolog, RSH, as well as other two monofunctional synthetases, RelP and RelQ, which respond to cell-wall-targeting antimicrobials (33). In this organism, (p)ppGpp does not bind RNAP, but instead interacts with HprT/Gmk (enzymes involved in GTP synthesis) and inhibits their function, which in turn leads to a decrease of the levels of GTP and subsequent activation of CodY (a transcriptional repressor) and decrease in rRNA synthesis (34).

Thus, the stringent response is a highly conserved starvation response in bacteria which allows for efficient reaction to diverse environmental starvation cues. It is one prime example on how bacteria can modify their metabolism and behavior to react to external pressures that come from their habitat.

1.2.3. The general stress response

Stressful environments can also induce entry into stationary phase. Upon detection of a stressor, bacteria can activate what is known as the general stress response (GSR). This is a global transcriptional response which allows for increased survival in the presence of multiple stressors.

The GSR is regulated by alternative sigma factors and can vary depending on the bacterial species.

In Gram-positive bacteria, SigB (sB) is the master regulator of the GSR, affecting the transcription of about 150 stress-related genes (35), and has been extensively studied in the model system Bacillus subtilis. The stress signals that control sB activity include both internal signals, such as the metabolic and energetic state of the cell (for instance, stationary phase entry signals

(36)), and external signals such as low pH levels, and ethanol or heat shock (37, 38).

In Gram-negative bacteria, specifically in most of the Proteobacteria, sS is responsible for the modulation of the GSR. It has been extensively studied in the model organism Escherichia coli

(a d-proteobacterium), where RpoS has been shown to respond to a wide array of diverse stressors, including low pH, heat shock, UV, osmotic stress and intracellular growth signals (11).

Not all Gram-negative bacteria encode for rpoS. Indeed, the GSR in the a-proteobacteria is instead tied to another alternative sigma factor: the extracytoplasmic function sigma factor EcfG

(also named sT) (39). Since the members of the a-proteobacteria class live in diverse environments

(see below and Appendix A), they activate the GSR in response to a wide array of stress signals, which include temperature shifts, desiccation, stationary phase, variation in pH levels, and UVs

(40–42).

1.3. THE a-PROTEOBACTERIA CLASS

Proteobacteria comprise the largest bacterial phylum. This phylum is divided into nine classes, including the aforementioned a-proteobacteria and the g-proteobacteria. The a- proteobacteria are one of the most diverse groups of bacteria and can be isolated from the most wide-ranging environments. For instance, the representative member of the Caulobacterales order, the asymmetrically dividing Caulobacter crescentus, was isolated from a pond in California (43); the aerobic anoxygenic phototroph (AAP, see Appendix A) Erythrobacter litoralis (of the

Sphingomonodales order) was found in the cold salt waters off the island of Texel, in the

Netherlands; the type species of the Rhizobium genus (of the order of the Rhizobiales), Rhizobium leguminosarum, instead grows in the roots of plants and in soil. Members of this class are also commonly isolated from the deep ocean floors (44) as well as from volcanic regions (45).

Furthermore, there is a large diversity in genome architecture and size within this class and even within the same order: some genomes, like that of Pelagibacter ubique (from the SAR11 clade) is one of the most compact genomes, with an average intergenic distance of 3 bp and a genome of

8 only 1.3 Mbp, making it the smallest known genome to date of this class (46); others, like

Mesorhizobium loti, are much larger, about 7 Mbp in size (47), all the while sharing the same order

– i.e. Rhizobiales – with Bartonella quintiana, whose genome size is only 1.7 Mbp (48).

The Alphaproteobacteria class also includes numerous unique pathogens. Among these are obligate intracellular pathogens like Rickettsia, which have a reduced genome (1.3 Mbp) due to permanent association with eukaryotic host cells, and the facultative intracellular pathogen

Brucella, which possesses two chromosomes (total of 3.3 Mbp). In fact, bacteria that are in perpetual contact with eukaryotic cells (such as the obligate Alphaproteobacterial symbionts,

Wolbachia) tend to have lost those genes that are no longer relevant to their survival, for instance biosynthesis genes for various metabolites. This is due to their nutritionally rich environment (i.e. the eukaryote host, cytoplasm) from which they can import vitamins and nutrients instead of having to expend energy in biosynthesis. On the other hand, those bacteria that live in nutrient- poor, changing environments are often more metabolically versatile, which enables them to adapt to more diverse nutritional conditions (49–51).

Thus, there is great variability within the alpha clade of proteobacteria, and they are uniquely useful model systems to study a variety of different morphological, genetic, environmental and evolutionary questions. The richness of the diverse environments they inhabit and the horde of distinctive challenges they face provides an outstanding reservoir of information to learn.

1.4. THE PATHOGEN BRUCELLA

Brucella are Gram-negative facultative intracellular pathogens, and members of the a- proteobacteria class. In 1887, David Bruce isolated the causative agent for a severe human disease

9 known as Malta fever, Mediterranean fever or Undulating Fever (now known as brucellosis), naming the pathogen Bacillus melitensis (now Brucella melitensis) (52). Subsequently, in 1897,

Bernhard Bang discovered the etiological agent responsible for abortion in cattle, and named it

Bacillus Bang (now Brucella abortus) (53). In fact, bacteria of the genus Brucella are the etiologic agents of brucellosis, which is among the most common zoonotic diseases worldwide (54, 55). It impacts human health (56) as well as the economy of affected regions as the most common practice to date to deal with an infected animal is to cull the entire herd (57). Brucella spp. can infect a range of wild and livestock animals (58), but have a relatively narrow host range and varying zoonotic potential (59, 60) and pathogenicity in humans. B. melitensis, for example, is primarily a sheep and goat pathogen and is often considered the most virulent species in humans (61). In contrast, human infections by the cattle pathogen, B. abortus, and the swine pathogen, B. suis

(biovars 1, 3, and 4), are less frequent and often less pathogenic, though clinical differences between these species are difficult to discern (62). Considering that genetic identity across the genus is 94–98% at the coding level (60, 63), differences in Brucella spp. host range, virulence, and zoonotic potential are notable. Comparative studies of genome content and the genetic requirements for growth in diverse environments may inform our understanding of differences in the physiology and infection biology of Brucella spp.

To date, there are 12 identified members of the Brucella family. They have been historically named according to the host they were originally discovered in and this usually reflects their preference in host species. The first three members identified were Brucella melitensis,

Brucella abortus and Brucella suis, which preferentially infect ovine, bovines and swine, respectively. These bacteria all form smooth colonies on agar plates, which indicates they form a full lipopolysaccharide (LPS) layer (see below), are highly virulent, and have zoonotic potential.

There are only two recognized rough species of Brucella (i.e. lacking the O-chain of their LPS):

Brucella canis (first described in 1966) (64) and Brucella ovis (discovered in 1953) (65), which were isolated from dogs and sheep, respectively. Other members of the Brucella clade include

Brucella neotomae (found in desert wood rats, 1957) (66), Brucella pinnipedialis (1996) (67) and

Brucella ceti (1994) (68, 69), both isolated from marine mammals, Brucella microti (found in voles, 2007) (70), Brucella inopinata (discovered in a breast implant, 2010) (71), Brucella papionis (in baboons, 2014) (72), and Brucella vulpis (isolated from a red fox, 2016) (73). While most Brucella were isolated from mammals, recent reports have identified novel Brucella spp. in amphibians (74, 75), thus increasing the range of possible hosts for this pathogen.

1.5. PATHOGENESIS

Brucella is a highly infective pathogen, its infectious dose varying based on the host and the route of entry, from tens to thousands (76) of bacteria. It can penetrate the human host via inhalation, ingestion, or direct contact with mucosa or damaged skin, is phagocytosed by professional (mostly dendritic cells (DCs) and macrophages) and non-professional phagocytes and then disseminates within the host (56, 77). Brucella infection usually progresses in three stages: an initial incubation stage, where there is no clinical evidence of infection and Brucella has passed the mucosal layer and entered the host; an acute phase, characterized by high numbers of Brucella disseminating through the host; a chronic phase, where the pathogen persists indefinitely, leading to organ damage and even death (78). In humans, the hallmark for brucellosis is the presence of an undulant fever, though this disease can present with a wide variety of symptoms (62).

Brucellosis is primarily an animal disease, and human to human transmission is a rare event

(59, 79). In the animal host, disease progression follows a different course than in the human

11 patient (80). It can infect the host via the gastrointestinal tract, sexual contact or inhalation. From the local lymph nodes, it spreads mainly through the reticuloendothelial system, often initially targeting organs like the kidneys and spleen, though the exact dissemination pattern strongly depends of the host. It also has strong tropism for male genital organs, mammary glands, and placental trophoblasts, in the case of a pregnant host. Indeed, abortion is a hallmark of animal brucellosis and the aborted fetus is an important source of contagions for livestock, wild-life and humans, although animals can also initially present with an undulant fever.

Brucella is described as a stealthy pathogen. This is because it is particularly apt at hiding from the immune system, particularly the innate immune response (81). Compared to other enteric pathogens, whose aggression is accompanied by strong clinical effects such as diarrhea and high aggressive fevers, Brucella only elicits a mild innate immune response, and, in fact, the first clinical and characterizing symptoms of brucellosis in humans is an undulant fever often accompanied by abdominal pain. Indeed, strong activation of the innate immune system involves pro-inflammatory cytokines (TNF-a, IFN-g, and IL-1b), and the presence of leukocytes

(especially neutrophils) at the site of inflammation, as well as in the stool and blood, which are not a characteristic of brucellosis. This furtive pathogen effectively counteracts immune activation by many means. For instance, it can interfere with DC maturation through action of Brucella-TIP protein 1 (Btp1), thus dampening immune response and hindering the production of TNF-a and

IL-12, and it does not bind the C3 component of the complement, which is another innate immune defense that attacks invading pathogens (78, 81–86).

Brucella also express an altered form of LPS, which is a pathogen-associated molecular pattern (PAMP) that is recognized by membrane-bound external toll-like receptor 4 (TLR4). LPS- from Gram-negative pathogens typically elicit a strong innate immune response, via activation of

12 the downstream pathways of TLR4. TLR4 triggers two main signaling pathways: the MyD88- dependent and the MyD88-indepentent pathways. The first leads to the activation of two transcription factors, NF-kB and AP-1, which have roles in the expression of pro-inflammatory cytokines. The MyD88-independent pathway activates the IRF3 transcription factor which induces production of type 1-interferons, such as IFN-a, which help regulate the activity of the immune system (87). Reduced TLR4 activation on the onset of Brucella infection means that there are only mild increases of pro-inflammatory cytokines like TNF-a and IL-1b, although some activation does occur at later stages of infection (82, 88, 89).

Furthermore, Brucella have evolved flagellin that lacks a domain that has been shown to elicit recognition by TLR5. TLR5 is another membrane-bound external TLR that specifically recognizes bacterial flagellin and activates NF-kB, AP-1, and IRF3 in response to the presence of this foreign protein (90, 91). Flagellin is the protein that constitutes the flagellar protofilament.

Brucella are described as a non-motile pathogens (61) and there has been no evidence of the presence of chemotaxis genes or motility, with the exception of a Brucella species that was isolated from a Pac-man frog (Ceratophrys ornata) (74). This is surprising as most Brucella species encode the whole host of flagellar genes (92). Of note, there are reports describing the expression of a flagella-like structure within a sheath, which has been linked to virulence in Brucella suis (93).

What the exact role of flagellin and flagellar encoding genes is in the context of brucellosis remains to be discerned.

1.6. THE INTRACELLULAR NICHE

Once Brucella is inside the host, it requires uptake by professional and non-professional phagocytes to establish infection (83). Only about 10% of Brucella survive phagocytosis (78, 94,

95). Nevertheless, this is an essential step in bacterial colonization.

Within the host cell, Brucella reside in what is known as the Brucella Containing Vacuole

(BCV) which will mature from early (eBCV) endosome to late endosome and eventually fuse to the lysosome, leading to a drastic drop in pH (96). It appears that the endosomal compartment physiologically matures, unperturbed by the presence of Brucella. Indeed, it has been shown that it initially displays early endosome markers (like Rab-5 and EEA-1), then late endosome markers

(such as CD63 and Rab7) and finally lysosomal markers, including LAMP1 and LAMP2, and bears the mark of a phagolysosome: pH levels of about 4-4.5. This happens within the first 8 hours post entry, with the lysosomal fusion occurring after the first 4 hours (97).

The lysosomal fusion is a fundamental step in Brucella host-cell infection. The drop in pH constitutes a signal for a set of virulence factors, most importantly the type IV secretion system

(T4SS) operon, virB, whose expression peaks at 4 hours post-infection. T4SS is a secretion system which pumps effectors in the host cytosol (98). These effectors help promote further BCV maturation to the replicative BCV (rBCV), which is established between 8 – 12 hours post infection. At this stage, the eBCV loses endosomal markers and gains ER-related ones, such as calreticulin and calnexin. The T4SS effectors are still largely uncharacterized, but they seem to allow the eBCV to interact with the ER exit sites (ERES), with COPII (which coats vesicles trafficked from the rough ER to the Golgi) and Sar1 (a small GTPase that control COPII-mediated trafficking), among others, thus promoting the formation of the rBCV (99). The ER-derived vacuole is now an ideal niche for Brucella to replicate within the host cell, even though there are reports of replication initiation occurring before the rBCV has been completely established (97).

As the infection continues, the rBCVs fuse together in vast membranous structures that contain large numbers of bacteria. These vacuoles, known as the autophagosome BCV (aBCV), do not actually display autophagosome markers (like LC3) but rather late endosome and lysosome ones (such as LAMP1 and acidification). Autophagy is a cellular process that allows for internal degradation of cytoplasmic contents, whether from foreign invaders or from damaged intracellular proteins or organelles and does not appear to be completely activated by Brucella. Rather, a subset of autophagy-related proteins (such as Becklin1 and ULK1) are required for aBCV formation.

Brucella can now egress from the cell (although this process is poorly understood) presumably by pirating the secretory functions of the autophagocytic pathway, and infect new cells. Of note, this process is non-lytic, and some Brucella maintain residence within the original host cell (100) while others egress and further disseminate within the host.

Within the host cell, Brucella is thus faced with a challenging and changing environment

(101). First of all, it encounters nutrient limitation, which is known to induce entry into stationary phase (102), until the rBCV is established. Furthermore, Brucella can be subject to a direct attack from reactive oxygen species (ROS) and reactive nitrogen species (RNS) (103), often associated to the respiratory burst (104, 105) (see below), a conserved host cell response usually aimed at attacking invading pathogens (106). Additionally, as mentioned above, Brucella have to withstand drastic drops in pH levels once the lysosome fuses to the vacuole, and they are further subject to attacks by antimicrobial peptides as well as nitrosative stress. Some of these stressors are described in more detail below.

1.6.1. Nutrient limitation within the BCV

Once inside the macrophage and before the establishment of the rBCV, Brucella experience nutrient limitation. As described above, nutrient limitation is one of the factors that can

15 lead to entry into stationary phase which then becomes an interesting physiologic state to study how Brucella may behave within the intracellular environment. Indeed, the similarities of the two, stationary phase and the intracellular niche, have been nicely reviewed by Martin Roop and colleagues (102). For instance, hfq in Brucella has been shown to confer resistance in late stationary phase in B. abortus and to play a role in in vivo BALB/c mouse infection studies (107), and there are a number of genes in Brucella that are dependent on Hfq function, including sodC

(which encodes for a superoxide dismutase, see below), that are important in Brucella in stationary phase (102). In 2006, the stringent response RelA/SpoT homolog rsh (see above), was shown to be important for virB expression and thus has an impact in virulence in Brucella (25), and in 2013

Hanna and colleagues studied the Rsh-dependent transcription profile in Brucella suis via microarray. Among others, they found that genes involved in stress adaptation (for instance sodC), protein folding, and chaperones were upregulated, whereas genes involved in amino acid and protein metabolism were largely downregulated. Expression of the transcriptional regulator hutC, which is a co-activator for virB, was also found to be upregulated in an Rsh-dependent manner

(108) .

1.6.2. The respiratory burst and oxidative stress

Professional phagocytes have evolved to fight off invading pathogens. One strong defense mechanism they employ is known as the respiratory burst (109). When phagocytes sense pathogen invasion, they attack by secreting various species of oxidants. These oxidants include ROS (110), reactive nitrogen species (RNS), and reactive chlorine species (RCS) (103, 111, 112). They can also be produced endogenously (i.e. through aerobic respiration, in the presence of free metals and erroneously oxidized non-respiratory flavoproteins) (106).

Whether encountered in the environment or produced during growth, bacteria must avoid accumulation of high levels of these oxidants (113). One important gene involved in protecting against ROS is superoxide dismutase (SOD), which was first characterized in 1969 (114). SOD is

•– a metalloenzyme (either Cu, Zn, Mn) that detoxifies the cells from the superoxide anion (O2 )

•– producing H2O2 (O2 + O2 + 2 H2O à H2O2 + O2). Hydrogen peroxide may react with metal ions such as iron(II) to produce a hydroxyl radical (OH•) (this is known as the Fenton reaction, Fe2+ +

3+ • H2O2 à Fe + OH ) (115). There are various enzymes that cells produce to decompose H2O2 into water and oxygen (2 H2O2 à 2 H2O + O2), namely catalases (116), glutathione peroxidases (117) and peroxiredoxins (118). In the case of RNS, the best characterized defense mechanisms include flavohemoglobins (Hmp) and flavorubredoxins (NorV). These enzymes detoxify NO• to nitrate (2

• – + + NO + 2 O2 + NADPH à 2 NO3 + NAD(P) + H ) in the case of Hmp (119), or to nitroxyl (3

• – + + NO + NADPH à NO + N2O + NAD(P) + H ), in the case of NorV (120).

Bacteria basally express scavenging systems as well as antioxidants to survive toxic oxidant accumulation. Among the various antioxidants, cysteine, methionine and glutathione are the most common. For instance, the thiol (-SH) group on cysteine’s side chain is a strong nucleophile, which reacts readily with electrophilic species, producing sulfenic acids (-SOH) in the presence of two-electron oxidants (such as peroxides) or thiyl radicals (RS•) if the electrophile donates a single electron. If the thiyl radical reacts with a hydroxyl radical, the product is the highly reactive sulfenic acid (-SOH), which can either form disulfide bonds with a proximal cysteine or be irreversibly oxidized to sulfinic (-SO2H) and sulfonic acid (-SO3H) (121). Methionine residues can also be oxidized to methionine sulfoxides (Met-O) and rarely, Met-O is further and irreversibly oxidized to methionine sulfone (122).

Bacteria can also protect themselves from the oxidative stress, for instance, by storing cysteine in LMW thiols, thus protecting against metal-catalyzed autooxidation, and by sequestering metal ions (such as Fe2+) in ferritins, bacterioferritins and Dps proteins, so as to reduce the chance of Fenton reactions taking place. Glutathione (GSH) is a low molecular weight

(LMW) thiol that contains a reactive sulfhydryl group. It acts as a redox buffer during oxidative stress and helps maintain a reduced cytoplasm. It is found in plants, animals, fungi, in most Gram- negative bacteria and in some Gram-positives, and is a tripeptide (cystine, glutamate and glycine).

Its two-step biosynthesis begins with fusion of cysteine to glutamate (catalyzed by the glutamate- cysteine ligase, GshA in Brucella) which produces g-glutamylcysteine. This is the rate-limiting step, and is followed by the addition of glycine, catalyzed by the glutathione synthetase (GshB).

Both steps are adenosine-triphosphate (ATP) dependent. GSH peroxidases can reduce peroxides to water by oxidizing two molecules of glutathione (2 GSH + H2O2 à GSSG + 2 H2O) as well as detoxify free radicals (GSH + R× à 0.5 GSSG + RH). Oxidized glutathione (GSSG) is once more returned to its reduced state a NADPH-dependent GSH reductase (Gor) (123–125). Of note, not all bacteria synthesize glutathione (106, 126, 127). For instance, in Mycobacteria, the predominant

LMW is mycothiol (MSH) and bacillithiol (BSH) is the most common LMW in Bacillus bacillithiol (128).

•– In the early endosome, transmembrane NADPH oxidase (NOX) will generate O2 and

H2O2 within the lumen of the vacuole. Furthermore, inducible nitric oxide synthases (iNOS) deliver RNS, such as NO• and peroxynitrite ONOO–. In some cell types, like neutrophils, myeloperoxidases can use the generated H2O2 to synthesize hypochlorous acid (HOCl), which is a two-electron oxidant that mainly targets proteins (106, 129, 130). Pathogens must thus find ways to survive the respiratory burst. Some employ numerous strategies to avoid the phagosome

18 altogether: for instance Yersinia pestis can disrupt signals that lead to the formation of the phagosome. Others, like Salmonella enterica, inhibit the production of ROS and RNS from within the phagocytic vacuole (131). When they do get bombarded by oxidants, pathogens can then employ various countermeasures to survive the assault, mainly by sensing the attack and increasing the basal levels of antioxidants they normally produce. Three master sensors are OxyR, PerR and

OhrR. OxyR is a transcriptional regulator belonging to the LuxR family, that was initially identified in Salmonella. It is conserved in Gram-negative and is present in some Gram-positive bacteria. It is mainly a peroxide sensor, thanks to a conserved cysteine residue that is oxidized in response to hydrogen peroxide (it can detect down to 1 µM of extracellular H2O2). Upon activation, it upregulates transcription of catalases and SOD, among other genes, in response to peroxide stress. In pathogens, its depletion leads to attenuated virulence. PerR also mostly responds to H2O2.

It is a transcriptional regulator of the Fur family that was discovered in Bacillus subtilis, and is mainly found in –but not limited to– Gram-positive bacteria. It is a zinc metalloenzyme that contains either Fe2+ or Mn2+ coordinated to histidine residues. Upon oxidation of the metal ion, repression of target genes is relieved in the case of PerR-Zn-Fe enzymes. In the case of PerR-Zn-

Mn, induction does not de-repress the PerR regulon, which includes catalases and other genes to defend against oxidative stress and metal homeostasis genes. OhrR is also a transcriptional repressor, belonging to the MarR family, but is only minimally activated by H2O2. Instead, it is best inactivated by organic hydroperoxides (R-OOH) and sodium hypochlorite (NaOCl), that interact with a conserved cysteine residue, which in turn relieves repression of target genes, such as glutathione peroxidase, as well as genes involved in quorum sensing and tyrosine metabolism

(132–134).

Pathogens that are unable to withstand or fight off oxidative stress are attenuated. For instance, the (p)ppGpp defective mutant in Staphylococcus aureus showed increase expression of

PerR, among other oxidative stress regulators, during stationary phase, and was more sensitive to hydrogen peroxide stress as well as ciprofloxacin- and tetracycline-mediated killing (135). In

Brucella abortus, SOD-deficient strains were found to be attenuated in both tissue culture and mouse infection models and pathogens developing resistance to RNS are a growing concern (136,

137).

1.6.3. Sulfur metabolism and pathogens

It is well established that sulfur metabolism plays a significant role in oxidant detoxification. Sulfur is essential for life and is incorporated into amino acids (cysteine and metabolism) as well as in essential cofactors (like coenzyme A (CoA), GSH, iron-sulfur clusters, and biotin). Different sulfur species can be imported into the bacterial cell. In the case of pathogens, inorganic sulfur is usually acquired as sulfate (SO2 –) and reduced first to sulfite (SO2 4 3

– ) then to sulfide (H2S), where it can be attached to organic molecules, such as O-acetylserine

(OAS) to produce cysteine, O-succinyl-homocysteine to produce L-homocysteine, and onto formaldehyde to produce methanethiol. If the bacteria encode a particular OASS, CysM, thiosulfate can also suffice as the sulfur source. Organic sulfur sources are also often extensively used by bacterial pathogens, such as cysteine, methionine and glutathione. Cysteine is a precursor in the biosynthesis of many sulfur-containing molecules, including GSH (see above) and methionine may also provide a reliable sulfur source if converted into cysteine (see below).

Furthermore, glutathione may be metabolized by bacterial cells to indirectly provide S. Indeed,

GSH is produced by eukaryotic cells (from 1-10 mM intracellularly and close to 40 µM in human plasma), and synthesis is induced upon oxidative conditions to protect the host, for instance in the

20 case of the respiratory burst. Many pathogens encode a g-glutamyl-transpeptidase (Ggt) that releases glutamate and g-cysteinyl-glycine that can then release cysteine via peptidase action (138).

Cysteine and methionine are considered essential amino acids in mammals, as they need to be acquired by diet and are not synthesized. Cysteine biosynthesis from serine is a two-step pathway (see also Chapter 3, Figure 6.16A); the rate-limiting first step is catalyzed by an O- acetylserine transferase (known as SAT or CysE, encoded by cysE in bacteria). CysE attaches an acetyl group activated with coenzyme A (acetyl-CoA) to serine, forming O-acetylserine (OAS).

Sulfide is then added to OAS by an acetylserine sulfhydrylase (OASS, encoded by cysK or cysM in bacteria) to form cysteine. The transsulfurylation pathway connects cysteine and methionine

(see also Chapter 4, Figure 6.30). Cysteine may be converted to cystathionine by a cystathionine- g-synthase (CGS, metB) then to homocysteine by a cystathionine-b-lyase (CBL, metC). This is known as the transsulfurylation pathway. Homocysteine is then converted to methionine by methionine synthases. The reverse pathway is also possible: methionine may be transformed to S- adenosylmethionine (SAM) by an SAM synthetase, then to homocysteine in a two-step reaction.

From homocysteine, a cystathionine-b-synthase (CBS) catalyzes the formation of cystathionine which is then substrate for a cystathionine-g-lyase (CGL) in the production of cysteine. These last two steps are part of what is known as the reverse-transsulfurylation pathway. Not all bacteria encode for the different enzymes involved in the transsulfurylation pathway (or the reverse process), thus modulating the different nutrients they can process and the different conditions they can grow in.

Defects in sulfur and cysteine metabolism can have effects on pathogen virulence. For instance, inhibitors targeting Salmonella typhimurium OASS have shown promising antimicrobial potential (139) and have also been tested in the treatment of Mycobacterium tuberculosis infection

(140). In Acinetobacter baumanii, functioning cysteine metabolism/sulfur assimilation pathways were found to be important in growth within the Galleria mellonella larvae-model of infection in a transposon-library experiment (141) and disruption of a cysteine metabolism transcriptional regulator (GigC) has been shown to play a role in virulence (142). In Staphylococcus aureus, defects in cysK have been shown to augment sensitivity to hydrogen peroxide stress and deletion of tcyP (encoding a transporter involved in sulfur acquisition) lead to competitive defects in murine models of infection (143, 144). Analysis of envelope proteins from Brucella abortus identified

CysK1 (see below) as an immunogenic candidate when the envelope proteome was probed with antisera from a patient (145), and CysK2 from Brucella abortus was later shown to elicit an immunological response in mice (146).

1.7. BRUCELLA OVIS, THE RUNT OF THE LITTER

An unusual member of the Brucella genus is Brucella ovis (147). It is an understudied outlier within this almost-monomorphic family of pathogens. One of the most easily observable characteristic that sets B. ovis apart from other Brucella (and that it shares only with B. canis, as mentioned above) is the fact that it forms rough colonies on plates. This phenotype is linked to its lipopolysaccharide (LPS) structure. Indeed, the envelope of Gram-negative bacteria can be decorated with this large molecule, which is constituted of three main parts: 1) lipid A, a phosphorylated glucosamine disaccharide bound to fatty acids, anchors the LPS to the cell outer membrane; 2) the core saccharide component, which latches directly onto lipid A; 3) the O-chain, which is a repetitive glycan polymer, is the outermost portion of the LPS. Colonies of bacteria with intact LPS are smooth in appearance, while colonies with their LPS lacking the O-chain don’t have a smooth colony texture, and are called “rough”. LPS can come in many flavors and serve

22 various purposes, from granting structural integrity to the cell envelope, to protection, to aid in surface attachment. It is also a potent endotoxin, eliciting a strong immune response through toll- like receptor 4 (TLR4), thus promoting secretion of pro-inflammatory cytokines, nitric oxide and eicosanoids (148, 149), see above. Rough mutants of naturally smooth Brucellae lose virulence, but both Brucella ovis and Brucella canis remain naturally virulent, even though they lack the O- chain of their LPS (98, 150–152). Since LPS is involved in cell integrity, variation in envelope structure can render different Brucella species more or less susceptible to environmental stressors, specifically membrane stress (see the work from Herrou et al. (153) for examples of Brucella sensitivity to membrane stressors).

B. ovis is also the only non-zoonotic member of this family, as there are no reported cases of B. ovis infections in humans, and it appears to have a narrower tropism than other members of the genus: indeed, it is mostly limited to sheep, although there have been reports of cases within the red deer populations in New Zealand (154). This allows experimental procedures to only warrant BSL-2 facilities and practices, compared to the BSL-3 requirements for other Brucella.

Furthermore, it seems to only enter the host through direct or venereal contact, and not through ingestion or inhalation. This limits its infective capacity drastically.

B. ovis also has a very specific tropism for the male genital tract in rams, where it causes epididymitis, orchitis and infertility. Compared to other Brucella, as mentioned above, B. ovis enters the mucosa, reaches the local lymph nodes then makes for the testes and epididymis, without affecting other organs. Indeed, for transmission to occur, the usual route needs an infected male to mate with a female and another healthy male to come along and mate with the ewe in the same estrus. Indeed, females tend to only suffer B. ovis infection transiently, and rarely experience abortion because of it (154–157).

Thus, there are many peculiarities concerning this outlier. At the genomic level, Brucella ovis shares quasi-genomic identity with other Brucella (94% nt identity with Brucella abortus), but has a highly degenerate genome. Indeed, about 244 of its genes are degenerated pseudogenes.

These genes include enzymes involved in various metabolic and virulence processes. For instance, the urease pathway is non-functional in Brucella ovis (158). The urea-degradation pathway releases CO2 and ammonia, and is utilized by various pathogens to survive the acidity of the stomach. This might explain why B. ovis cannot be transmitted via ingestion (80). Another example of an affected pathway is that of erythritol oxidation (159). This sugar is thought to be one of the preferred carbon sources for Brucella, is found in the placentas of goats, cows, and pigs, and has been linked to virulence in Brucella (160), possibly linking the strong tropism Brucella have for placental trophoblasts in animals and account for the limited capacity of B. ovis to induce abortion in ewes. Of note, a large deletion of 15 kb in the region that encodes for glycosyl transferases (which are in involved in LPS biosynthesis) accounts for the rough colony phenotype of B. ovis (80). Finally, this accumulation of pseudogenes and non-functional pathways also leads to its heightened sensitivity to partial CO2 pressures (pCO2) (161, 162), as discussed in detail in

Chapter 2 and Chapter 4.

Thus, Brucella lead a dangerous lifestyle (77, 82), and are constantly challenged by a dangerous environment (163). Studying how Brucella sense and react to these changing and challenging environments is key to understanding their physiology and infection, and eventually better fight this stealthy pathogen.

1.8. BRUCELLA CULTIVATION AND CO2: A BRIEF HISTORY

As outlined above, Brucella ovis is an interesting outlier in the genus. One of its defining characteristics is its inability to grow in laboratory conditions without CO2 supplementation.

Carbon dioxide is an incredibly important molecule that permeates, directly or indirectly, into the metabolism, signaling, pH homeostasis, and in short, in the life of all organisms. Understanding how the environmental levels of CO2 affect Brucella ovis remains an important question and is an old historical problem.

Bang and Stribolt pioneered the study of B. abortus growth in axenic culture, which was important in establishing its identity as the etiologic agent of contagious abortion in cattle (53). Of particular note in their original study was the growth pattern of B. abortus in solid shake tubes, in which they observed bacteria in two bands below the surface of nutrient agar. Based on a series of experiments using different atmospheres of CO2 and O2, Bang concluded that oxygen and not carbon dioxide was the gaseous factor that governed growth, though he noted significant variability in growth zones in shake tubes (53). In the years following the discovery of Bang and

Stribolt, reliable axenic cultivation of B. abortus from clinical specimens remained a major challenge (164).

Huddleson conducted a series of growth experiments in 1921, demonstrating that increasing the partial pressure of CO2 (pCO2) stimulated growth of B. abortus clinical isolates

(165). A follow-up investigation provided additional evidence that CO2 was required to support metabolism of B. abortus, and that the effect of CO2 on growth was not a result of pH changes in the medium (166). G.S. Wilson later presented evidence that both oxygen and CO2 per se are required for B. abortus growth, and that bicarbonate cannot substitute for CO2 in the growth medium (167).

To assess the metabolic fate of CO2, J.B. Wilson and colleagues grew multiple B. abortus

14 strains in the presence of CO2 and measured incorporation of carbon into amino acids and nucleotides (168–170). Carbon-14 largely pooled in pyrimidines and in glycine: 90% of the carbon-14 detected in the amino acid pool was in glycine (170). Amino acids, purines, pyrimidines, various organic acids and cell hydrolysates were not able to substitute for CO2 in their experiments

14 (171). In the case of B. abortus, the incorporation of CO2 into glycine and pyrimidines occurred at similar ratios in all strains tested, even those that did not require increased pCO2 for growth

(170). From these data, the authors concluded that fixation of CO2 is a fundamental property of B. abortus strains.

1.8.1. Natural variation in the requirement for CO2 supplementation

When investigating the cause of Malta fever, Bruce noted that Micrococcus (Brucella) melitensis appeared as colonies “on the surface” of nutrient agar in an air incubator (52). This stands in contrast with Brucella abortus isolates from cattle that did not grow on agar surfaces (53) without CO2 supplementation (165). Indeed, variation in atmospheric requirements for growth of these bacteria clouded the fact that these early investigators were actually studying closely-related species (172). Pioneering work by Alice Evans united the etiologic agents of Malta fever in humans and contagious abortion in cattle into a single group (173) that she classified under the genus

Brucella (174). By 1930, it was established that certain Brucella isolates required CO2 supplementation for axenic cultivation, while others could be cultivated without increased pCO2

(175). In the case of B. abortus, it was well known by the 1920’s that the requirement for increased pCO2 for growth was often spontaneously lost after a period of cultivation in the laboratory (166,

176). Buddle reported a similar phenomenon in an early description of Brucella ovis isolated from rams in New Zealand (177).

What, then, occurs when a Brucella isolate spontaneously loses the requirement for increased pCO2 to grow? Early descriptions of “crops” of colonies that grew from clinical specimens on solid media after several days of incubation (166) are reminiscent of spontaneous genetic mutants that arise when bacteria are cultivated under selective conditions. Smith attempted to isolate genetic revertants of lab-adapted strains derived from these crops by exposing the cells to guinea pig tissue or re-infecting cows (166). However, he failed to identify strains that regained a requirement for CO2 supplementation. Marr and Wilson later demonstrated that B. abortus

-10 spontaneously loses its requirement for CO2 at a rate of 3 ´ 10 per generation (178), supporting a model whereby this phenotype is likely a result of mutation at a single site.

1.9. CO2 METABOLISM AND CARBOXYLASES

Upon noting the differences in requirement for CO2 supplementation between different

Brucella species, one might wonder why CO2 is important in bacterial growth. Actually, carbon dioxide is a biologically important molecule at all levels, and it plays into various aspects of the life of all organisms. CO2 is first and foremost the prime carbon source for autotrophs, and different organisms have evolved strategies to acquire carbon by fixing CO2, though the relevance carbon dioxide acquisition is not limited to autotrophy. CO2 fixation is dependent on the activity of carboxylases, and there are at least six different CO2-fixation pathways that have been described.

The importance of carboxylases is linked to the biological role of CO2 in the growth and survival of all life. The first class of carboxylases are the autotrophic carboxylases fix CO2, thus deriving biomass that all organisms use (directly or indirectly). The most prominent member of this class of carboxylases is, in fact, ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO). The second group of carboxylases are those involved in assimilatory pathways. Indeed, various

27 biological reactions require addition of a carboxyl group as a functional group to different substrates for additional transformations to occur, especially under anaerobic conditions. For instance, in aiding incomplete b-oxidation, where propionate assimilation requires propionyl-CoA to be transformed into (2S)-methylmalonyl-CoA by a biotin-dependent carboxylase since b- oxidation of fatty acids cannot metabolize the short odd-numbered fatty acid, allowing for the production of succinyl-CoA synthesis from (2S)-methylmalonyl-CoA, an intermediate of the tricarboxylic acid cycle (TCA cycle). Biosynthetic carboxylases belong to the third group, which includes carboxylases involved in the biosynthesis of specific types of molecules, for instance carboxylases involved in fatty acid metabolism. During the extension of the fatty acid chain, malonyl-CoA is obtained by addition of acetyl-CoA. A biotin-dependent carboxylase uses incorporated CO2 to activate the acetyl-CoA molecule that will then be used to extend the chain.

Carboxylases are also important in redox balancing and are involved in removing excess reducing equivalents (i.e. NAD(P)H). CO2 thus functions as an electron sink, for instance in Rhodobacter capsulatus, a purple non-sulfur bacteria of the a-proteobacteria class, where RubisCO activity becomes essential if the bacteria are grown photoheterotrophically in the presence of highly reduced substrates like butyrate. Finally, the last group of carboxylases are those involved in anaplerotic reactions. When cells are actively growing and central carbon metabolism is active with many intermediates being drained from the TCA cycle as biosynthetic building blocks, the cycle needs its intermediates replenished to continue to function. Anaplerotic carboxylases like biotin-dependent pyruvate carboxylases and phosphoenolpyruvate carboxylases transform pyruvate or phosphoenolpyruvate, respectively, to oxaloacetate (179–181).

CO2 is also the product of aerobic respiration and other biological reactions. It is present in the Earth’s atmosphere at about 0.04% (though levels are increasing, https://www.co2.earth/)

28 but can be found in higher concentrations in different habitats, for instance within human cells and tissues. Indeed, mammalian cells have evolved up to 13 bicarbonate transporters to remove dissolved CO2.

A lot of important carboxylation reactions require an essential cofactor: biotin. Brucella carries the whole set of genes required for biotin synthesis. Biotin, also known as vitamin B7, is a heterocyclic sulfur-containing cofactor. All biotin-dependent enzymes transfer CO2 via a two-step reaction, which is also important in assimilation of carbon dioxide (see previous chapter). In the first step, biotin is carboxylated, binding CO2. In the second step, the biotin cofactor releases the carbon dioxide which is either released or supplied to a substrate. There are three classes of biotin- dependent enzymes: carboxylases, decarboxylases and transcarboxylases. The latter (class III) is the only class that does not use free CO2. Class II biotin-dependent enzymes free CO2 from a substrate and release it. Finally, Class I enzymes are biotin-dependent carboxylases. They add free

CO2 to an acceptor, which distinguishes between families of this class (182, 183).

1.10. CARBONIC ANHYDRASES

As detailed in Chapter 2, the molecular basis for the difference in in CO2 requirement between Brucella lineages is the activity of a carbonic anhydrase. I will thus briefly introduce carbonic anhydrases and their biological relevance. As mentioned above, CO2 in the atmosphere is not particularly abundant, and natural hydration to bicarbonate (which is the biologically relevant form of CO2) occurs to slowly on a metabolic level. Carbonic anhydrases are enzymes that intervene in acquiring atmospheric (gaseous) CO2 -which can freely pass through membranes-

– and trap it within cells in its hydrated ionic form, bicarbonate HCO3 .

Carbonic anhydrases were first purified 1933 from cow erythrocytes (184). They belong to a superfamily of metalloenzymes and are now known to present in all kingdoms of life. They

– + catalyze the reversible hydration of carbon dioxide to bicarbonate (CO2 + H2O « HCO3 + H ), although some have been shown to catalyze additional reactions, such as the hydration of carbonyl sulfide (COS) (185) and carbon disulfide (CS2) (186) in bacteria, with important roles for their survival. Although different carbonic anhydrases display different kinetics, some are among the

8 -1 fastest known enzymes, kcat/KM > 10 M s (187).

A divalent cation in the active site is required for catalytic activity, which is bound in a tetrahedral geometry, with one of the coordination bonds supplied by water. The most common is zinc (Zn2+), which can be found lodged in the active site of all carbonic anhydrases, though the physiologic ion of choice varies between classes. To this day, eight distinct families have been identified (188): a-, b-, g-, d- , z-, h-, q-, and i-carbonic anhydrases. There is very little sequence and structural similarity between the different classes as carbonic anhydrases seem to have evolved independently and are a great example of convergent evolution. The enzymes from the different families differ in the preferred coordinated cation, in the 4D structure (i.e. some work as monomers, either as dimers, trimers, or higher) and in the specific organisms they may be isolated from.

The carbonic anhydrases that we will discuss in depth in Chapter 2 and in Chapter 4 are of the beta family. The beta family is found in Gram-negative and Gram-positive bacteria as well as in fungi, some Archaea, in chloroplasts of mono- and di-cotyledons, and in algae. They function as dimers or as multiples of dimers and also find Zn2+ in the active site. They can further be subdivided into class I or class II. In class I b-carbonic anhydrases, the active site is in an open conformation and shows only one histidine and two cysteines binding the metal ion. In class II,

30 the water molecule is replaced by an aspartate residue, which causes the active site to be in a closed conformation. These carbonic anhydrases are not active unless the pH increases above 8.3, where a conserved asparagine can then interact with the aspartate, freeing the active site and allowing a water molecule to enter (187–190)

1.10.1. Biological function of CO2 and carbonic anhydrases

High CO2 levels can damage to cells and tissues, and rapid conversion to bicarbonate allows for maintenance of pH homeostasis, ion secretion (191). Furthermore, carbonic anhydrases play an important function in biosynthetic pathways and in various anaplerotic reactions; for instance, they are involved in gluconeogenesis, lipogenesis (pyruvate carboxylase and acetyl CoA carboxylase), amino acid biosynthesis (through the activity of pyruvate carboxylase), and ureagenesis (192) (carbamyl phosphate synthetase I), pyrimide synthesis (carbamyl phosphate synthetase II). Then also play roles in CO2 sensing and respiration (193) and in mammals they have been implicated in electrolyte secretion, bone resorption (194), calcification, in urinary acidification and bicarbonate reabsorption (in the kidneys) (195) and associated diseases (191).

They also play an important role in tumorigenesis and tumor hypoxia. In algae and plants, carbonic anhydrases have also been shown to participate in photosynthesis and related biosynthetic reactions, and where, interestingly, CA expression can be induced by low CO2 coupled to the presence of light (196–198). Furthermore, in diatoms, the delta and zeta families of CAs are involved in carbon fixation (199, 200) and h-carbonic anhydrases seem to play a role in de novo purine/pyrimidine biosynthesis, possibly presenting a novel target for anti-malarial agents (201–

203).

Finally, carbonic anhydrases have been linked to virulence in diverse pathogens (187). In

Staphylococcus aureus (a commensal and opportunistic pathogen Gram-positive pathogen of the

Firmicutes phylum), a recent report showed that MpsAB, a proposed sodium bicarbonate cotransporter, to be important in for growth in atmospheric CO2 conditions as well as play a role in virulence (204). In the case of Salmonella enterica serovar Typhimurium (a member of the g- proteobacteria and causative agent of typhoid fever), disruption of a putative carbonic anhydrase

(mig-5) which is expressed within the host did not have an effect in an tissue culture infection assay, but was attenuated in a mouse model (205). In Escherichia coli (another commensal and opportunistic pathogen), the requirement for carbonic anhydrase activity was extensively studied examining the well-established E. coli carbonic anhydrases, Can and CynT (206). Lastly, Brucella appears to have three carbonic anhydrases, two b-carbonic anhydrase (BcaA and BcaB) and a g- carbonic anhydrase (RicA). Studies on the role of these CAs’ roles in metabolism, physiology, pathogenesis and inhibition have been conducted over the past eleven years and will also be discussed in further in Chapter 2 and Chapter 4 in the context of the facultative intracellular pathogen Brucella ovis (161, 162, 207–210).

1.11. CONCLUSIONS

There are many shifts in the environment that bacteria need to deal with, in order to survive and thrive. The facultative intracellular pathogen lifestyle of Brucella is a fascinating model to query for this aspect. The challenge it faces within a host, specifically within a host cell, are as unique as the genetic determinants Brucella employs to withstand its natural habitat. Shifting it away from its natural niche and into a laboratory setting, the fastidiousness of Brucella ovis with regards to CO2 is a bell that points to a particular form of adaptation of this understudied outlier within the Brucella family. We were curious to understand this phenotype more clearly and its link to other Brucella lineages (see Chapter 2 and Chapter 4). Within the intracellular space, within

32 a vacuole, B. ovis is constantly stressed, persistently fighting off attacks of different natures. As such, understanding how it survives within the BCV, how this relates to stationary phase, and the metabolic requirements that underlie this line of defense is an interesting and pertinent avenue to explore, and our related findings are detailed in Chapter 3 and further discussed in Chapter 4.

2. A CARBONIC ANHYDRASE PSEUDOGENE SENSITIZES

SELECT BRUCELLA LINEAGES TO LOW CO2 TENSION

2.1. PREFACE

The contents of this chapter were modified and adapted from its published form:

Varesio LM, Willett JW, Fiebig A, Crosson S., Journal of Bacteriology (2019)

DOI: 10.1128/JB.00509-19.

Data Sets may be found online.

2.2. INTRODUCTION

The requirement for CO2 supplementation for Brucella ovis growth in laboratory conditions is an interesting feature of this member of the Brucella family. We sought to understand the genetic determinants that were responsible for this phenotype and link our discoveries to the

Brucella family as a whole. Understanding how environmental levels of carbon dioxide affect growth of Brucella ovis and what advantage or disadvantage there may be in limited CO2 assimilation is another angle at which to probe its biology and pathogenesis.

2.3. RESULTS

2.3.1. A forward genetic selection identifies B. ovis mutant strains capable of growth in an unsupplemented atmosphere

Brucella ovis requires 5% CO2 supplementation to grow axenically (80). We sought to identify genes linked to this growth requirement, and thus developed a forward genetic selection to identify spontaneous B. ovis mutants that grow in a standard, unsupplemented atmosphere

(0.04% CO2). Specifically, we inoculated wild-type B. ovis in Brucella broth (BB) and allowed the cultures to shake in an air incubator. After several days, some cultures became visibly turbid, providing evidence for growth (see Figure 6.2A and Materials and Methods). Twenty-nine mutant strains were isolated from five independent selections. We inoculated all mutants into fresh broth and confirmed that they indeed grew without CO2 supplementation (Table 7.1).

Genomic DNA from 16 of these spontaneous mutants was purified and sequenced along with DNA from the wild-type B. ovis parent. Each mutant carried a single nucleotide deletion within a six-nucleotide interval at the 3’ end of a b-carbonic anhydrase gene (locus tag

BOV_RS08635; bcaABOV) (Figure 6.3A and Table 7.2). These deletions in bcaABOV frameshifted the coding sequence of the BcaABOV C-terminus and increased the length of the predicted gene product from 207 to 214 residues (Figure 6.3A-C). The 16 sequenced mutants clustered into 4 groups based on the site of the frameshift mutation; the corresponding alleles are named bcaA1BOV-

4BOV (Figure 6.3C). Nucleotide deletions in bcaABOV were the only mutations that were common to all mutant strains and missing from the parent (Table 7.2).

Similar bcaABOV mutations in B. ovis strains that grow without CO2 supplementation have been recently described by Perez-Etayo et al. (162), though they have used the name carbonic anhydrase II (CAII) for the gene locus and product. For the purposes of this manuscript, we have chosen to use the bacterial genetic naming guidelines of Demerec et al. (211), and thus refer to this gene as bcaABOV. We conclude that cultivation of wild-type B. ovis in a standard atmosphere selects for a frameshift mutation at the 3’ end of the b-carbonic anhydrase gene, bcaABOV. This

35 frameshift increases the length of the predicted protein and completely changes the sequence of the last 32-33 residues of BcaABOV. Based on the fact that we successfully selected mutants in standard air after several days of incubation, we presume that wild-type B. ovis can replicate (albeit very slowly) in an unsupplemented atmosphere.

2.3.2. A bcaA frameshift enables growth in an unsupplemented atmosphere

To directly test the hypothesis that bcaABOV frameshift mutations enable growth in an unsupplemented atmosphere, we built a strain in which wild-type bcaABOV was replaced with the frameshifted bcaA1BOV allele. We also built a strain in which bcaABOV was deleted in-frame

(DbcaA) (Figure 6.3D). We measured growth of these strains and wild-type B. ovis in 5% CO2, and in standard atmospheric CO2 levels (approx. 0.04%). All three strains grew in the presence of

5% CO2, indicating that bcaABOV is not required for growth in a CO2-supplemented atmosphere.

When these strains were incubated with 0.04% CO2, only the bcaA1BOV allele replacement strain showed measurable growth. Notably, the bcaA1BOV strain grew significantly faster (i.e. had a shorter doubling time) than wild-type and DbcaA strains in the presence of 5% CO2 (Figure 6.2B and Figure 6.3E). From these data, we conclude that frameshift mutations at the 3’ end of bcaA1BOV enable robust growth of B. ovis in an unsupplemented atmosphere and result in faster growth than wild-type in the presence of 5% CO2.

2.3.3. bcaA frameshift mutations are the sole genetic mechanism by which B. ovis growth is spontaneously rescued under low CO2 tension

To test whether mutations in genes other than bcaABOV enable B. ovis growth in an unsupplemented atmosphere, we repeated the forward genetic selection described above starting with the DbcaA strain. We incubated DbcaA in BB at 37°C for over two weeks in an air incubator

36 and never observed turbid cultures. This experiment was repeated 4 times, with approximately 30 inoculated tubes each time, and growth was never observed. This stands in contrast with our initial selection experiment, in which multiple tubes inoculated with wild-type B. ovis became turbid over this timescale. As an alternative forward genetic approach, we generated a pool of B. ovis Tn-

Himar mutants with insertions at over 5 ´ 104 unique sites (Table 7.3). This experiment was aimed at identifying insertional mutations that enable growth in a standard atmosphere. We incubated aliquots of this pool in BB for over two days in an air incubator and did not observe growth. Thus, none of the transposon insertions in the pool enabled B. ovis growth in the absence of CO2 on this short timescale. From these data, we conclude that single nucleotide deletions at the 3’ end of bcaABOV, which result in a frameshifted gene product, are the sole genetic mechanism by which B. ovis spontaneously acquires the ability to grow without added CO2 on this experimental timescale.

2.3.4. The frameshifted alleles, bcaA1BOV-4BOV, are dominant over the wild-type bcaABOV allele

We next transformed wild-type B. ovis with low-copy replicating plasmids, from which we

++ ++ expressed either the wild-type (bcaABOV ) or frameshifted (bcaA1BOV ) alleles from an IPTG- inducible (lac) promoter (Plac) (Figure 6.3F). A strain carrying an empty vector (EV) was used as

++ a control. The EV and the bcaABOV strains (with 1 mM IPTG) grew comparably in 5% CO2 and did not grow when shifted to a standard atmosphere (0.04% CO2) (Figure 6.2C and Figure 6.3G).

Thus, overexpressing wild-type bcaABOV from a plasmid is not sufficient to enable growth in a

++ standard atmosphere. The bcaA1BOV strain grew without added CO2, demonstrating that the frameshifted allele is dominant and sufficient to enable B. ovis growth in 0.04% CO2. Moreover,

++ ++ the strain expressing bcaA1BOV grew significantly faster in 5% CO2 than the bcaABOV and EV

37 strains, providing additional evidence that bcaA1BOV enhances growth rate in broth culture (Figure

6.2C and Figure 6.3G).

To confirm that all four classes of frameshifted bcaA alleles identified in our forward genetic selection (bcaA1BOV-4BOV) were sufficient to enable growth in a standard atmosphere, we expressed these alleles from a replicating plasmid in a standard air incubator (0.04% CO2) and compared growth to the B. ovis EV strain. As expected, all four alleles enabled growth in 0.04%

CO2 (Figure 6.2D). Thus, different single nucleotide deletions at the 3’ end of bcaABOV (Figure

6.3C) result in the same phenotype: growth of B. ovis in a standard atmosphere. Given that the four alleles phenocopy each other, we continued our study with the bcaA1BOV allele. This was the most common isolate from our forward genetic selection and is the most conserved bcaA allele in the entire Brucella clade (see below).

2.3.5. Expression of E. coli b-carbonic anhydrases is sufficient to enable B. ovis growth in a standard atmosphere

bcaABOV is predicted to encode a b-class carbonic anhydrase. These enzymes catalyze the reversible hydration of CO2 to bicarbonate, an important anabolic substrate. Carbonic anhydrases have been extensively studied (187, 190) in diverse species(191, 212, 213), including Brucella suis

1330 (207, 208, 214). The thirty-three C-terminal residues of wild-type B. ovis BcaABOV differ from B. suis 1330 BcaA. However, the sequence of the frameshifted mutant B. ovis BcaA1BOV protein is identical to B. suis BcaABSU1330 (locus tag: BR_RS08400; also known as Bs1330CAII (162) or bsCAII (214)) with the exception of one residue (Figure 6.4A). The B. ovis bcaA1BOV mutation thus results in a frameshifted gene product with a C-terminus that matches B. suis BcaABSU1330, as well as BcaA orthologs in other Brucella species, which we discuss below.

B. suis 1330 BcaABSU1330 is reported to be an active b-carbonic anhydrase (207, 214). Thus, we reasoned that the selected frameshift mutations at the 3’ end of B. ovis bcaABOV restored the protein to an active form that can hydrate CO2 at the low partial CO2 pressure of a standard atmosphere, thereby enabling growth. Following this hypothesis, we tested whether heterologous expression of known active Escherichia coli MG1655 b-class carbonic anhydrases can (locus tag: b0126) or cynT (locus tag: b0339) (206) was sufficient to support growth of the wild-type and

DbcaA B. ovis strains in a standard atmosphere (Figure 6.5A). Induction of can or cynT expression from a low-copy plasmid (RK2-Plac) with 1mM IPTG had no effect on growth rate in 5% CO2

(Figure 6.5B), but enabled growth of wild-type and DbcaA B. ovis in an unsupplemented atmosphere (0.04% CO2) (Figure 6.5C). We conclude that expression of E. coli can or cynT

(Figure 6.4B) is sufficient to enable growth of WT and DbcaA B. ovis strains under the low CO2 tension of a standard atmosphere and, therefore, that carbonic anhydrase activity in and of itself is sufficient to allow wild-type or DbcaA B. ovis to grow in 0.04% CO2.

2.3.6. A comparative analysis of bcaA sequences and the CO2 growth requirement in the genus Brucella

As outlined in the introduction, B. ovis and the majority of B. abortus strains require CO2 supplementation for axenic cultivation (215). To assess the extent to which variation in bcaA is linked to this physiologic trait, we compared the sequences of 773 bcaA orthologs from the genus

Brucella (see Data Set 1). B. ovis and the majority of B. abortus encode BcaA variants that significantly deviate from the consensus sequence for the genus (Figure 6.6A).

All 17 B. ovis bcaABOV sequences in our dataset harbor an additional guanosine 522 nucleotides after the start codon (G522) relative to other Brucella species (Figure 6.3), which

39 results in a corresponding frameshifted BcaABOV C-terminus (Figure 6.3 and Figure 6.6). The B. ovis sequences we analyzed were from strains isolated from sheep in Europe, South America,

Australia, and New Zealand over the course of several decades (Data Set 1). Thus, this single nucleotide insertion in bcaABOV is apparently a defining genetic characteristic of B. ovis. As highlighted in Figure 6.3A-C and Figure 6.6B, single nucleotide deletion mutations that enable

B. ovis growth in a standard atmosphere (bcaA1BOV-4BOV) abolish this frameshift and restore the

BcaA C-terminus to the consensus of the Brucella genus.

We further analyzed bcaA from 374 B. abortus strains and observed two major sequence classes: 1) the majority of sequences (278) encode a bcaA pseudogene (B. abortus** in Figure

6.6), and 2) the remainder encode full-length proteins (B. abortus* in Figure 6.6) that completely or almost completely match consensus (see below). The bcaA pseudogene (bcaABAB) sequences all share the same C339 insertion that results in a frameshift and premature stop codon (Figure 6.7A, cluster 1). The bcaA sequences from the remaining 96 B. abortus strains, encoding a full-length

BcaA protein, grouped into seven clusters. The largest of these clusters contains sequences lacking the C339 insertion (52 out of 374 strains), resulting in a consensus BcaA gene product (Figure 6.6 and Figure 6.7A, cluster 2). In the other B. abortus bcaA clusters the consensus reading frame was restored by either a single nucleotide deletion or two-nucleotide insertions within 12 base pairs

(downstream) of the C339 frameshift site (Figure 6.7A). We posit that the bcaABAB pseudogene is the ancestral allele in B. abortus, and that insertion/deletion (i.e. indel) mutations near nucleotide

339 have been selected in different B. abortus strains/biovars to resurrect functional BcaA variants that enable growth under low partial CO2 pressure.

2.3.7. Functional B. abortus bcaA alleles are rapidly selected from a bcaABAB pseudogene under CO2 limitation

We next assayed the function of two B. abortus bcaA alleles (bcaABAB and bcaABAB2308) in

B. ovis (Figure 6.7B). Specifically, we replaced wild-type B. ovis bcaABOV with either the bcaABAB pseudogene or the near-consensus bcaABAB2308 allele (cluster 7, Figure 6.7A), and measured growth in a standard unsupplemented atmosphere (0.04% CO2). Replacing bcaABOV with bcaABAB2308 enabled growth of B. ovis in 0.04% CO2 (Figure 6.7C), consistent with Perez-Etayo et al. (162). Replacing bcaABOV with the frameshifted bcaABAB pseudogene did not enable B. ovis growth in a standard atmosphere (Figure 6.7C). We conclude that bcaABAB is inactive in vivo and that the majority of Brucella abortus isolates cannot grow in an unsupplemented atmosphere, which is consistent with historical observations outlined in our introduction (53, 165, 166, 176).

Conversely, approximately one-quarter of sequenced B. abortus strains harbor a copy of bcaA in which the frameshift mutation at C339 has apparently been remediated (Figure 6.7A) to restore expression of a full-length protein. Given these results, we hypothesized that functional variants of B. abortus bcaABAB could be experimentally selected in Brucella cells cultivated under standard atmospheric conditions, i.e. low pCO2.

To test this hypothesis, we used the B. ovis strain in which we replaced bcaABOV with the non-functional B. abortus bcaABAB pseudogene (carrying the C339 insertion) and selected for mutants that could grow in an unsupplemented atmosphere. Four independent experiments yielded spontaneous mutants that grew to high density without CO2 supplementation within several days.

We sequenced the bcaA locus of 5 independent mutants from these selections and identified three types of mutations (bcaA1BAB -3BAB, Figure 6.7D) that remediated the C339 frameshift and restored the full-length, consensus BcaA coding frame. All bcaA variants that we selected experimentally are present in sequenced isolates of B. abortus. We cataloged information of the 374 sequenced B. abortus isolates, including assigned biovars (when available) from PATRIC (Table 7.4). B.

41 abortus biovars 5 and 6 have been described as CO2 independent (215) and, as predicted, all sequenced isolates of these biovars possess consensus (i.e. full-length) bcaA alleles. Biovars 1, 3 and 4 are described as primarily CO2 dependent, which is consistent with the fact that between 70 and 95% of sequenced isolates from these biovars carry the bcaABAB pseudogene. B. abortus biovar

2 is categorized as CO2 dependent (215), yet 5 of 18 sequenced isolates possess a consensus functional bcaA allele; biovar 7 is classified as CO2 independent, though one of five sequenced isolates carries the bcaABAB pseudogene. We conclude that the identity of bcaA alleles in the different B. abortus biovars is largely consistent with the CO2 dependence/independence cataloged by Alton et al. (215).

Our analysis shows that the most common allele of bcaA in B. abortus is the frameshifted bcaABAB pseudogene. Thus, like B. ovis, many wild B. abortus populations apparently harbor a non-functional, frameshifted bcaA pseudogene that renders the bacterium unable to grow without

CO2 supplementation. Also, like B. ovis, functional (consensus) bcaA alleles can be easily selected from bcaABAB by incubating the bacterium in a standard atmosphere for several days.

2.3.8. CO2 limitation triggers a large-scale transcriptional response in wild-type B. ovis, while transcription in the bcaA1BOV strain is insensitive to CO2 shifts

Our group and others (162) have shown that bcaA is a specific genetic determinant of growth under low pCO2. We have further shown that B. ovis and most B. abortus strains harbor non-functional bcaA pseudogenes containing distinct single-base insertion mutations that result in a frameshifted gene product. These insertional mutations underlie the long-noted CO2 requirement for axenic cultivation of these species (see Chapter 1). Given these results, we sought to better understand the physiological consequences of CO2 limitation in B. ovis, a species that requires

CO2 for axenic cultivation. Specifically, we defined transcriptional changes in B. ovis ATCC

25840 at the genome scale in response a downshift in pCO2.

To measure the transcriptional response to CO2 change, we cultivated B. ovis wild-type and bcaA1BOV strains in a 5% CO2 atmosphere. A subset of these cultures was then shifted to a standard atmosphere (0.04% CO2). After a controlled incubation period in this condition (see

Figure 6.8A and Materials and methods) we harvested cells, extracted RNA from all samples, and prepared libraries for sequencing. Shifting wild-type B. ovis from 5% to 0.04% CO2 slowed growth (Figure 6.9A), and induced large-scale changes in gene expression: 382 genes were significantly upregulated and 442 genes were significantly downregulated upon a downshift in pCO2 (log2 fold change > |1|; false-discovery rate (FDR) p-value £ 0.001) (Data Set 2 and Figure

6.9B). In contrast, shifting a B. ovis strain harboring the bcaA1BOV allele from 5% to 0.04% CO2 had no effect on growth (Figure 6.9C) or gene expression using the same significance threshold

(Data Set 2 and Figure 6.8B). In fact, a comparison of wild-type B. ovis cultivated at 5% CO2 to bcaA1BOV cultivated in 0.04% CO2 revealed no differentially expressed genes. As expected from these results, the overall transcriptional profile of wild-type B. ovis cultivated in 5% CO2 is highly

2 correlated with the B. ovis bcaA1BOV strain cultivated under either 5% CO2 or 0.04% CO2 (R =

2 0.995, 5% CO2 and R = 0.993, 0.04% CO2) (Figure 6.8C and Figure 6.9D). We conclude that wild-type B. ovis is acutely sensitive to downshifts in levels of atmospheric CO2, and that presence of bcaA1BOV – the consensus allele in the Brucella clade – renders B. ovis insensitive to CO2 downshifts. Notably, our data show that transcription of bcaABOV itself is not induced by CO2 limitation (Figure 6.9E).

2.3.9. CO2 limitation induces a stringent-like starvation response and activates transcription of the virB type IV secretion gene cluster in wild-type B. ovis

We next assigned each of the genes up- or down-regulated upon pCO2 downshift in wild- type B. ovis to Kyoto Encyclopedia of Genes and Genomes (KEGG) categories. Of the 824 genes that had a log2 fold change > |1| and an FDR p-value < 0.001, 447 could be classified into one or more KEGG categories (Data Set 3). Only categories where 3 or more genes were assigned were considered in our pathway analysis (Figure 6.10). Genes involved in translation, including those encoding ribosomal proteins, ribosomal RNA, transfer RNA, and aminoacyl-tRNA synthetases, are largely downregulated upon shift from 5% to 0.04% CO2 (Data Set 2, 3 and Figure 6.10). In addition, transcript levels for elongation factors P, Tu and G are highly reduced upon CO2 downshift. This transcriptional profile, which reflects reduced protein synthesis, is consistent with nutrient limitation and activation of the stringent response (23, 29, 30, 216). In addition, multiple respiratory, energy metabolism, transport, and catabolic genes are downregulated upon CO2 downshift (Data Set 2, 3 and Figure 6.10). Consistent with the fact that a starvation, or stringent- like response, is induced by CO2 downshift, we observe a lag phase in WT B. ovis growth upon re-introduction to a 5% CO2 environment (Figure 6.9A).

Expression of genes implicated in response to stress, nutrient limitation, or stationary phase, including polyphosphate kinases (ppk1, ppk2) (217–219), small DUF1127-family proteins

(BOV_RS04430, BOV_RS13510, BOV_RS13790, BOV_RS13950) (220), alternative sigma factors rpoH (221) and ecfG/rpoE1 (222) is activated upon CO2 downshift. Furthermore, Entner-

Doudoroff/pentose phosphate pathway genes (zwf, pgl, edd) are transcriptionally activated, as are genes implicated in Brucella spp. virulence and infection including the virB type IV secretion gene cluster, the urease cluster, and several annotated flagellar genes (Figure 6.10 & Figure 6.11 and

Data Set 2). Although B. ovis is urease negative and possesses a degraded urease gene cluster (80), transcription of the ure genes (and pseudogenes) is activated upon CO2 downshift. Among the

44 most highly activated gene sets in low CO2 are biotin biosynthesis genes (Figure 6.10 & Figure

6.11 and Data Set 2, 3). Biotin is a cofactor involved in carboxylation reactions that use CO2 as a substrate. Concordant with the result that cells remodel transcription to enhance carboxylation reactions, we also observed that transcription of pyruvate carboxylase and several biotin- dependent acetyl-CoA carboxylases and carboxyl transferases is significantly enhanced (Figure

6.11 and Data Set 2).

2.3.10. Pseudogene sequences, including bcaABOV, are highly conserved across all B. ovis isolates

All 17 sequenced B. ovis isolates (Data Set 1) carry the same single nucleotide insertion in bcaABOV, yielding a frameshifted gene that apparently encodes a non-functional carbonic anhydrase protein. From these data, we have concluded that bcaABOV is a pseudogene. Since pseudogenes are predicted to be under neutral selection (223–225), we might expect that bcaABOV should accumulate mutations at a higher frequency than functional genes. Actually, there is evidence that pseudogenes are rapidly deleted from bacterial genomes due to either toxic side effects of expression, or the energetic burden of maintaining the gene (226). However, we observe no other mutations within bcaABOV across all sequenced B. ovis strains, which have been isolated from regions across the globe over the past 60 years. This suggests either that the nucleotide insertion mutation at the 3’ end of bcaABOV occurred very recently in the evolutionary history of the B. ovis lineage, or that selective pressures maintain this pseudogene, raising the possibility that this gene sequence may have an additional function. It is known that bcaABOV does not contribute to B. ovis infection in a mouse model of disease (162), but we have shown that this gene is transcribed (GEO accession GSE130678 and Figure 6.9E) and thus could contribute other functions. To assess if the conservation of bcaABOV allele in B. ovis differs from other

45 pseudogenes, we defined the frequency of mutations and sequence divergence across a larger group of both annotated pseudogenes and functional genes, both within in B. ovis and compared to B. abortus strain 2308. Our aim was to use this gene set to empirically define pseudogene divergence in B. ovis, and to use this set as a point of comparison for bcaABOV.

We quantified polymorphisms in three distinct sets of genes; 1) urease genes, 2) the VirB type IV secretion system (T4SS), and 3) B. ovis pseudogenes described by Tsolis et al. (80) (Table

7.5). B. ovis is urease negative (80, 227) and the ure gene clusters show features of degradation

(80) with intact genes and pseudogenes that are presumably both under neutral selection as the pathway is non-functional. B. abortus, on the other hand, is urease positive and has an intact ure gene cluster (228). The T4SS provided us with a set of genes that are functional and intact in both species (229–231). The additional B. ovis pseudogenes included in our analysis are multiple erythritol catabolic genes and cytochrome oxidase genes, which have functional, full-length orthologs in B. abortus (Table 7.5). When assessing polymorphisms, each gap (insertion or deletion) was considered as a single event, regardless of its length.

We observed strikingly low divergence within B. ovis isolates. In other words, the sequences of all B. ovis pseudogenes in our analyzed set are highly conserved across a geographically diverse group of strains isolated between 1959 and 2011. Only 4 pseudogenes — ureT (BOV_RS06530), norB (BOV_RS11505), ccoO (BOV_RS01915) and pckA

(BOV_RS09880) — exhibit polymorphisms between B. ovis strains, and each is polymorphic at only one site (Table 7.3). Similarly, sequence differences are rare when comparing the ure genes or the T4SS genes within the 17 B. ovis strains. In short, the level of polymorphism we observe comparing the sequenced B. ovis isolates is not significantly different between a functional group of genes (virB) and a non-functional gene cluster containing pseudogenes (ure).

We then sought to compare these genes to B. abortus ATCC 2308. Overall, annotated pseudogenes have the highest average number polymorphisms per base pair between B. ovis strains and B. abortus. As expected, genes in the functional T4SS (virB) locus have the lowest average number of differences per base pair and a bias towards synonymous mutations (Figure 6.12). B. ovis does not produce functional urease and the ure genes, accordingly, have a higher number of mutations compared to B. abortus. We do observe that larger or out-of-frame deletions are more prevalent in the ure genes than the virB cluster (Table 7.5) consistent with a model in which the ure genes are under relaxed selective pressure. Though pseudogenes are, on average, more highly diverged between B. ovis and B. abortus (Figure 6.12), our analysis shows that many other pseudogene sequences besides bcaABOV are also entirely conserved in across sequenced B. ovis isolates (Table 7.5).

Thus, we cannot conclude that the conservation of bcaABOV across B. ovis isolates is due to purifying selection as a result of BcaABOV performing some other function in the cell. Rather, a more likely explanation for our results is that the loss of function mutation resulting in a B. ovis bcaABOV pseudogene occurred very recently in the evolutionary history of an ancestral B. ovis strain that now commonly infects sheep across the globe.

2.4. DISCUSSION

2.4.1. Brucella cultivation and the resurrection of a carbonic anhydrase pseudogene

During the course of transmission and infection, Brucella spp. must endure numerous shifts in the chemical state of the environment (59, 96), including changes in steady-state levels of CO2.

In the aqueous environment of the cell, CO2 can spontaneously hydrate to form bicarbonate, an important substrate for microbial growth (179). The importance of bicarbonate to life is evidenced

47 by the fact that at least 7 different families of carbonic anhydrases (188, 190, 191) have evolved across all kingdoms of life. These enzymes function to enhance the rate of CO2 hydration to produce bicarbonate. Brucella spp. encode at least three carbonic anhydrases, but our data and recently published data (162) provide evidence that a b-class carbonic anhydrase encoded by bcaA is specifically required for axenic growth under the low partial CO2 pressure (pCO2) of a standard atmosphere in B. ovis and B. abortus.

Almost all Brucella spp. can be cultivated without adding CO2, but the majority of B. abortus strains and all wild B. ovis strains have a strict requirement for CO2 supplementation for axenic cultivation. This requirement confounded early efforts to cultivate B. abortus from cattle

(53, 165), and the knowledge that B. abortus strains require added CO2 enabled early efforts to cultivate B. ovis from sheep tissue (65, 177). Our analysis of over 700 sequenced Brucella isolates of multiple species provides evidence that single nucleotide insertions in bcaA result in frameshifts that lead to non-functional BcaA protein in B. abortus and B. ovis. Both of these species carry distinct insertion mutations in bcaA (Figure 6.3 & Figure 6.7and Data Set 1) that underlie the strict CO2 requirement for growth outside of an animal host. Functional, reverted alleles of these frameshifted B. abortus and B. ovis bcaA pseudogenes are easily selected by incubating cells in broth under standard atmospheric conditions for several days. This is consistent with historical reports of loss of the CO2 requirement after repeated passages of B. abortus and B. ovis in air (166,

177), and is conceptually in line with a recent report in E. coli showing functional resurrection of a pseudogene involved in iron acquisition when cells are cultivated under selective conditions

(232).

The fact that B. ovis and B. abortus lineages have independent insertion/frameshift mutations in bcaA that are apparently fixed in populations of each species suggests there may be

48 an adaptive advantage to loss of bcaA function. In the case of B. ovis, our data show that mutants harboring the functional bcaA1BOV allele grow significantly faster than wild-type at high partial

CO2 pressure (Figure 6.3E and G), which is what Brucella spp. encounter in mammalian host tissues. The spleen colonization phenotype of B. ovis strains carrying a functional bcaA1BOV allele does not differ from wild-type B. ovis in a mouse model of infection (162), but it is known that a slow growth rate can be advantageous for bacterial pathogens to maintain long-term infections in certain tissues (233, 234). Thus, it is possible that loss of bcaA function enhances Brucella fitness or persistence in tissues of natural (i.e. non-rodent) animal hosts by slowing growth. We note that

B. ovis is almost entirely restricted to male reproductive tissue and is reported to be transmitted only through direct or venereal contact, via a transiently infected ewe (57, 157, 177). The rate of oxygen uptake and the partial pressure of CO2 are higher in testes than in other tissues (235, 236), and it is therefore conceivable that B. ovis has evolved a lifestyle in which carbonic anhydrases are no longer required. B. abortus has a tropism for bovine reproductive tissue in males and females, and it may be also the case that loss of bcaA function influences its rate of growth and its persistence in these tissues, though this hypothesis remains to be tested.

There is no known Brucella reservoir outside of animals, but it is well established that presence of bacteria on aborted tissue is one of the main routes of transmission for B. abortus (77).

A functional BcaA would seem to be advantageous for CO2 assimilation when B. abortus is residing on such tissue, though the extraordinarily high titer of respiring brucellae on an aborted fetus (237) may lead to a high enough local pCO2 that BcaA is unnecessary for growth in this context. Alternatively, Brucella growth may be arrested during the window in which it is outside its preferred host tissue. Growth impairment upon CO2 limitation may, in fact, serve as an important signal for the bacterium during the transmission cycle. We do not currently understand

49 why loss of bcaA function is restricted to the B. ovis and B. abortus lineages, or if there is a particular feature(s) relating to CO2 metabolism or transmission/infection biology that sets these species apart from other members of the genus. It is possible that all Brucella species, as intracellular animal pathogens, will eventually lose bcaA function.

2.4.2. CO2 limitation elicits a starvation and virulence gene expression response

14 As outlined in the introduction, carbon-14 from supplied CO2 was recovered primarily in pyrimidines and glycine in B. abortus (170, 238). Our efforts to bypass the B. ovis CO2 requirement by supplementing media with amino acids, nucleotides, and bicarbonate were not successful, just as earlier efforts to bypass the B. abortus CO2 requirement by adding bicarbonate to the medium were unsuccessful (167). Thus, it seems likely that Brucella spp. lack a functional bicarbonate transporter. Our transcriptomic data provide clear evidence that shifting B. ovis to a standard atmosphere induces a starvation response that has features of the stringent response (23, 30, 239).

This response is evidenced by the downregulation of transcription and translation related genes as well as genes involved in respiration and central metabolism (Figure 6.10). Among the most significantly activated set of genes upon CO2 downshift is the virB type IV secretion system, suggesting that cells initiate a gene expression program that could influence virulence when they encounter a CO2 limited environment. Growth and transcription in a B. ovis strain with a functionally-restored bcaA gene (i.e. bcaA1BOV) was completely insensitive to a shift in pCO2. We propose that loss-of-function frameshift mutations in bcaA yield strains that can more readily sense changes in pCO2 in the host. Beyond slowing the growth of B. ovis under the high partial CO2 pressure of the host environment, it is possible that loss of bcaA function also endows cells with an ability to more acutely detect changes in environmental CO2 levels, for example when a

Brucella cell is ejected from its animal host. Whether this change in the CO2 detection threshold is advantageous in natural infection and transmission contexts remains to be determined.

2.5. MATERIALS AND METHODS

Bacterial strains and growth conditions

The bacterial strains used in this study are listed in Data Set 4. Brucella ovis was grown on Schaedler agar (Difco Laboratories) or on Tryptic Soy Agar (Difco Laboratories) plates supplemented with 5% defibrinated sheep blood (Quad Five) (SBA or TSBA respectively) at 37°C with 5% CO2 when required. Liquid cultures were grown in Brucella Broth (BB) (Difco

-1 Laboratories) at 37°C with 5% CO2 when required. 50 µg ml of kanamycin (kan) was added when required. Expression strains were supplemented with 1 mM isopropy b-D-1- thiogalactopyranoside (IPTG) (GoldBio). All studies on Brucella ovis were performed following

Biosafety Level 2 (BSL2) protocols.

Escherichia coli strains used for cloning were grown on lysogeny broth (LB) (Fisher

Bioreagents) at 37°C and on LB plates with 1.5% agar (Fisher Bioreagents). Kan was added to a final concentration of 50 µg ml-1. For conjugation, the diaminopimelic acid (DAP) auxotrophic E. coli strain WM3064 was grown with a final concentration of 300 µM DAP (Sigma-Aldrich).

Plasmid and Strain Construction

For the construction of allele replacement strains, external primers overlapping regions approximately 500 bp upstream or downstream flanking target genes of interest were PCR amplified with KOD Xtreme Hot Start Polymerase (Novagen) using either Brucella ovis, Brucella ovis mutant or Brucella abortus genomic DNA as a template. For deletion strains, internal primers

51 were designed so that the 4-6 amino acids at the N-terminus and the C-terminus of the gene product would remain in frame to minimize polar effects. For strains with point or small mutations, the target gene was amplified using internal primers carrying the mutation of interest. The various fragments obtained were then joined by overlap extension PCR (OE-PCR) to generate null or mutant alleles using KOD. Insertion of fragments into the suicide plasmid pNPTS138 was attained either by restriction enzyme (RE) digestion and T4 ligase ligation or by Gibson assembly (New

England Biolabs).

To build gene overexpression strains, genomic DNA from Escherichia coli MG1655,

Brucella ovis, Brucella abortus or derivatives thereof was used as templates. Primers amplifying the gene of interest starting with the start codon and including the stop codon were amplified by

PCR with KOD. Fragment insertion into the lac-inducible, low-copy pSRK plasmid was attained either by RE digestion and T4 ligation or by Gibson assembly. All insertion sequences into plasmids were confirmed by Sanger sequencing. Sequence-confirmed plasmids were transformed into E. coli Top10 (for plasmid maintenance), purified (ThermoFisher Scientific), and transformed into E. coli WM3064 (William Metcalf, University of Illinois) for conjugation into Brucella ovis.

In the case of replicating plasmids, selection on SBA or TSBA kan plates (without DAP) allowed for isolation of single B. ovis clones containing the plasmid of interest. In the case of the pNPTS138 plasmid used to conduct gene replacement, cells from mating were first selected on SBA or TSBA kan plates. pNPTS138 also carries the sacB gene for counterselection. Strains harboring pNPTS138 integrants were outgrown in BB and spread on TSBA (or SBA) plates containing 5%

(w/v) sucrose. This permitted selection of single clones in which a second recombination event that excised the plasmid had occurred. Clones were then screened by PCR with GoTaq® Green polymerase and the PCR products were Sanger sequenced (University of Chicago Comprehensive

Cancer Center, DNA Sequencing & Genotyping Facility) to confirm the sequence of the gene deletion or replacement. See Data Set 4 for primers, strains and plasmids.

DNA extraction, amplification and quantification

Genomic DNA was extracted following a standard guanidinium thiocyanate protocol.

Briefly, strains were struck on SBA (or TSBA) agar plates and colonies were picked and cultivated in BB overnight. 1 ml of culture was spun at 12000 rpm for 20 s and the pellet washed with 0.5 ml of Phosphate-Buffered Saline (PBS). Pellet was resuspended in 0.1 ml of TE buffer pH 8.0 (10mM

Tris-HCl pH 8.0 (Fisher Bioreagents); 1 mM EDTA (Fisher Bioreagents)). TE buffer included ribonuclease A (1 µl ml-1). 0.5 ml of GES lysis solution (5 M guanidinium thiocyanate (Fisher

Bioreagents), 0.5 M EDTA pH 8.0; 0.5% v/v sarkosyl) was added. Following a 15 min incubation at 60 °C, 0.25 ml of cold 7.5 M ammonium acetate (Fisher Bioreagents) were added. After 10 min on ice, 0.5 ml of chloroform (Fisher Bioreagents) were added and samples were vortexed and centrifuged. The aqueous phase was mixed with 0.54 volumes of cold isopropanol (Fisher

Bioreagents) and incubated at room temperature for 15 min before centrifugation. The pellet was washed three times in 70% ethanol before resuspending in TE buffer + RNAseA, at which point the concentration was determined spectrophotometrically.

Forward genetic selection for mutants that grow without added CO2

To select for spontaneous mutants that grow without addition of 5% CO2 to the atmosphere, wild-type B. ovis ATCC 25840 or B. ovis DbcaA strains were inoculated into BB at a final optical density (600 nm; OD600) of ~0.3 then placed in a shaker (200 rpm) under standard atmospheric conditions (0.04% CO2). Cultures were incubated at 37°C and spectrophotometrically monitored for evidence of growth every 24 hours (Figure 6.2A). A total of 29 independent spontaneous

53 mutants were selected that acquired the ability to grow without added CO2. These mutant isolates

-3 -5 -7 -9 were back diluted to OD600=1.5 ´ 10 , 1.5 ´ 10 , 1.5 ´ 10 , 1.5 ´ 10 and grown in 0.04% CO2 to confirm that the isolates indeed grew without the addition of 5% CO2 (result from inoculum of

1.5 ´10-5 shown in Table 7.1). Samples were confirmed to B. ovis by PCR amplification of a

Brucella-specific gene (BOV_RS05580).

Whole Genome DNA Sequencing (WGS)

Genomic DNA (gDNA) from the parent B. ovis strain, which requires 5% CO2 to grow, and gDNA from 16 independent spontaneous mutants that grow without added CO2 (evolved from the parent stock) was purified using the standard guanidinium thiocyanate-based procedure described above. DNA sequencing libraries were prepared from randomly sheared DNA and sequenced using the standard Illumina protocol (Illumina HiSeq 4000; single end 50 bp reads).

Reads were mapped to the Brucella ovis ATCC 25840 genome (chromosome 1 and chromosome

2 RefSeq accession numbers NC_009505 and NC_009504, respectively) and polymorphisms were identified using breseq (240) (https://github.com/barricklab/breseq). Raw DNA sequencing reads for all strains are available in the NCBI Sequence Read Archive (SRA), through BioProject accession PRJNA540707.

Growth Curves

Strains were struck on SBA or TSBA plates and allowed to grow for ~48 hours. Initial inoculum into BB ranged from OD600 of 0.015 to 0.03. Antibiotics for plasmid maintenance and

IPTG for expression induction were added when needed at the start of the growth experiment.

Growth of two to three separate tubes as technical replicates per sample was monitored

54 spectrophotometrically at 600 nm (OD600). Each growth curve was measured at least three times independently.

Sequence alignments

Genome sequences of Brucella isolates were downloaded from PATRIC

(https://www.patricbrc.org/) (241). The amino acid sequence was aligned using the multiple alignment tool in Geneious 10.0.9 (www.geneious.com): global alignment with free end gaps,

Blosum 62 scoring matrix, gap open penalty of 12 and gap extension penalty of 3. To cluster the

Brucella abortus bcaA alleles, nucleotide sequences were aligned using Geneious multiple alignment, global alignment with free end gaps specifically, cost matrix set at 65% similarity, gap open penalty of 12, gap extension penalty of 3 and 2 refinement iterations.

RNA sample preparation and RNA sequencing

Samples for RNA sequencing were prepared as follows (see also Figure 6.8): six independent cultures each of wild-type B. ovis or B. ovis bcaA1BOV strain were inoculated into BB and allowed to grow to saturation overnight at 37°C in a roller in the presence of 5% CO2. Cells were back diluted to an OD600 of 0.05 and monitored until they reached OD600 of 0.1. Three tubes of B. ovis bcaA1BOV were transferred to a roller in a standard air incubator (i.e. 0.04% CO2), while three tubes of bcaA1BOV remained in the 5% CO2 incubator. Both sets of bcaA1BOV cultures were harvested after an additional 2.5 hours of growth. Cells from three tubes of wild-type B. ovis were harvested at this same time, while the last three tubes of wild-type B. ovis were transferred to the standard air incubator and rolled for an additional 2.5 hours. Wild-type B. ovis cultures showed no signs of growth over this period in air (0.04% CO2). Cells from the remaining three wild-type B. ovis tubes incubated in the atmospheric (0.04% CO2) incubator were then harvested.

RNA was prepped from all harvested cell samples as follows: each individual culture sample was aliquoted into six 1.5 ml Eppendorf tubes (for a total of 9 ml) and spun for 60 s at

14,500 ´ g. Pelleted cells were immediately re-suspended in 1 ml of Trizol (Ambion, Life

Technologies, ThermoFisher Scientific) and stored at -80 °C. Samples were then thawed at 65 °C for 10 minutes. Once thawed, 200 µl of cold chloroform was added, cells were vortexed for 15 s and incubated for 5 min at room temperature. Samples were then centrifuged at 4 °C (17000 ´ g) for 4 min and 500 µl of 100% cold isopropanol was added to the clear supernatant in fresh tube, then frozen for at least 1hr at -80 °C. Samples were centrifuged at 4 °C (17000 ´ g) for 30min, the supernatant was removed, and 1 ml of 70% ethanol was added to wash the pellet. After an additional 5 min centrifuge (at 4 °C, 17000 ´ g), residual ethanol was removed and the RNA pellet was resuspended in RNAse free water (50 µl). RNA was treated with DNAse and further purified with RNA purification kit (Qiagen). Concentrations were determined spectrophotometrically, and

RNA quality was initially assessed by running 1 µl of each sample on a Tris/Borate/EDTA (TBE)

2% agarose gel.

Ribosomal RNA was depleted from the sample using the Gram-negative bacteria Ribo-

Zero rRNA Removal Kit (Illumina-Epicentre). RNA-seq libraries were prepared with an Illumina

TruSeq stranded RNA kit according to manufacturer's instructions. The libraries were sequenced on an Illumina HiSeq 4000. Sequencing reads were deposited in the NCBI GEO database and are available under accession GSE130678.

RNA sequencing data analysis

Reads were mapped to the Brucella ovis ATCC 25840 genome (chromosome I and chromosome II RefSeq accession numbers NC_009505 and NC_009504, respectively) using CLC

Genomics Workbench 11.0 (https://www.qiagenbioinformatics.com); mismatch cost: 2; insertion cost: 3; deletion cost: 3; length fraction; 0.8, similarity fraction: 0.8. Two samples were extreme outliers: wild-type B. ovis (5% CO2) replicate 3 and B. ovis bcaA1BOV (0.04% CO2) replicate 1, based on principle component analysis of the raw expression values. This was likely due to RNA sample degradation in these samples during library preparation. Accordingly, these samples were not included in our differential expression analysis. Reads Per Kilobase per Million (RPKM),

Transcripts Per Million (TPM), and Counts Per Million (CPM) values for single samples as well as paired differential expressions – with p-values, False Discovery Rate (FDR) and Bonferroni corrections – are presented in Data Set 1. Significance thresholds were arbitrarily set at FDR p- value < 0.001 and fold change > |2|. KEGG analysis of significantly regulated genes in wild-type

B. ovis in low (0.04%) versus high (5%) CO2 conditions was performed using KEGG pathways

(242) (https://www.genome.jp/kegg/). KEGG ontology annotations were assigned to these genes.

This annotated gene list was submitted to KEGG Mapper–Search Pathways, querying against

Brucella ovis ATCC 25840 (bov). Pathways that had 2 or fewer genes were removed. We removed from consideration pathways where the number of upregulated versus downregulated genes was not significantly different. Specifically, pathways were removed if the number of the upregulated genes divided by the total number of genes assigned to that pathway was between 0.4 and 0.6.

Construction and mapping of B. ovis ATCC 25840 Tn-Himar mutant library

B. ovis ATCC 25840 was struck on an SBA plate and incubated for 48h before mating with

E. coli APA752 (WM3064 donor strain carrying pKMW3 mariner transposon vector library)

(243). E. coli APA752 was incubated overnight prior to mating (1 ml was thawed and inoculated into 25 ml of LB + DAP + kan). 10 ml of donor strain were pelleted and mixed with pellet of B. ovis ATCC 25840 scraped off an SBA plate at a 1:10 ratio, and resuspended in a total of 250 µl of

BB. 50 µl of the conjugation mixture were spotted on SBA + DAP plates and left to incubate overnight. Spots were then harvested and resuspended in 5 ml of BB. Cells were diluted to OD600 of 0.032 and 500 µl were plated on each of twelve 150 mm SBA + kan plates. Colonies grew in approximately three days and were harvested and resuspended in 250 ml of BB + kan at a final

OD600 of 0.2. Cells were allowed to double 2-3 times before being frozen in 15% glycerol. For mapping, the library was thawed and the cells were washed once in PBS before pelleting for DNA extraction. The Illumina sequencing library to map Tn-Himar insertion sites was built following the protocol of Wetmore et al. (243) and as previously described (153, 244) Briefly, adapters were added to the genomic library by PCR (30 min at 95 °C and 15 min at 70 °C) using Mod2_TS_Univ and Mod2_TruSeq primers. DNA was then sheared to obtain fragments approximately 300 to 500 bp in length. Samples were then size selected, end repaired, ligated to a custom adapter, and cleaned before 150 bp single end sequencing by the University of Chicago Functional Genomics

Facility. See Table 7.4 and Data Set 5 for list of sequencing statistics and unhit genes, respectively.

Statistical analysis

In bar graphs, error bars represent standard deviation (unless otherwise indicated) of replicates from at least three independent experiments (two or more technical replicates were used for each individual experiment). Strains that failed to grow are indicated as “no growth”. **** indicates significance of p<0.0001, *** indicates significance of p<0.001, ** is p<0.01; ns = non- significant, calculated using one-way ANOVA followed by Tukey’s test using GraphPad Prism version 8.0.2. For the scatter plot, R2 values were calculated by implementing linear regression in

GraphPad Prism. For heat map, z-scores were calculated as follows: for each gene, the CPM value was multiplied by the mean of CPM values for that gene across all samples. The resulting value

58 was divided by the standard deviation of the CPM values of that gene across all samples. The heat map was built using Java Tree View version 1.1.6r4 (http://jtreeview.sourceforge.net).

2.6. SUMMARY

Brucella are intracellular pathogens that cause a disease known as brucellosis. Though the genus is highly monomorphic at the genetic level, species have animal host preferences and some defining physiologic characteristics. Of note is the requirement for CO2 supplementation to cultivate particular species, which confounded early efforts to isolate B. abortus from diseased cattle. Differences in the capacity of Brucella species to assimilate CO2 are determined by mutations in the carbonic anhydrase gene, bcaA. Ancestral single nucleotide insertions in bcaA have resulted in frameshifted pseudogenes in B. abortus and B. ovis lineages, which underlie their inability to grow under the low CO2 tension of a standard atmosphere. Incubation of wild-type B. ovis in air selects for mutations that “rescue” a functional bcaA reading frame, which enables growth under low CO2 and enhances growth rate in high CO2. Accordingly, we show that heterologous expression of functional E. coli carbonic anhydrases enables B. ovis growth in air.

Growth of B. ovis is acutely sensitive to a reduction in CO2 tension, while frame-rescued B. ovis mutants are insensitive to CO2 shifts. B. ovis initiates a gene expression program upon CO2 downshift that resembles the stringent response and results in transcriptional activation of its type

IV secretion system. Our study provides evidence that loss-of-function insertion mutations in bcaA sensitize the response of B. ovis and B. abortus to reduced CO2 tension relative to other Brucella lineages. CO2-dependent starvation and virulence gene expression programs in these species may influence persistence or transmission in natural hosts.

3. BRUCELLA OVIS CYSTEINE BIOSYNTHESIS

CONTRIBUTES TO PEROXIDE STRESS SURVIVAL AND

FITNESS IN THE INTRACELLULAR NICHE

3.1. PREFACE

The contents of this chapter were modified and adapted from its published form:

Varesio LM, Fiebig A, Crosson S. Infection and Immunity (2021).

10.1128/IAI.00808-20.

Data Sets may be found online.

3.2. INTRODUCTION

Brucella spp. are intracellular pathogens that have numerous mechanisms to contend with host-generated stressors and exploit host resources for growth. Within the host, brucellae are subject to nutrient limitation (102), phagosomal acidification (245), and direct attack from reactive oxygen and reactive nitrogen species (103) originating from the host-derived respiratory burst

(104, 105). Dozens of genes involved in oxidative stress responses, acid stress responses, nutrient assimilation, and respiration have been implicated in the biology of Brucella infection (101). More recent studies have defined a role for the general stress response pathway in mitigation of multiple chemical stressors in vitro and in maintenance of chronic infection in vivo (222, 246). However, relatively little is known about the mechanisms Brucella spp. use to adapt to stresses encountered in axenic cultures during stationary phase. The study of stationary phase has the potential to inform

60 the discovery of genes that influence infection, intracellular replication, and survival (247) as there are postulated parallels between stationary phase physiology and the physiologic state of Brucella in the intracellular niche (102).

3.3. RESULTS

3.3.1. B. ovis cysE Tn-himar mutant strains have a fitness defect in stationary phase

We inoculated »1.5 × 109 B. ovis RB Tn-himar strains (see Materials and methods and

(244)) into Brucella Broth in triplicate and collected samples at intervals throughout the growth curve: 0.05, 0.12, 0.9 and 2.4 OD600 (corresponding to early logarithmic, logarithmic, late logarithmic, and stationary phase). Barcodes were PCR amplified, sequenced, and tallied as previously described (243) to assess the relative abundance of each mutant strain in each sample.

Our analysis yielded composite fitness scores for 2638 of 3391 annotated genes in B. ovis (Data

Set 1). Data for 118 mutants that exceeded a t-like test significance threshold ≥ 4 are presented in

Figure 6.13 (see Materials and methods). We observed the largest relative fitness score changes at OD600 = 2.4 (i.e., stationary phase) in this dataset.

To more rigorously assess mutants with fitness values that varied as a function of growth phase, we further filtered the genes to include only those with a fitness score ³ |1| in at least one timepoint (see Materials and methods). We hierarchically clustered the 64 genes that passed this cutoff (Figure 6.14A, Data Set 1) and divided these clustered genes into four groups that displayed different fitness patterns throughout the growth curve (Figure 6.14B). Mutations in group 1 genes resulted in no fitness defect during exponential growth, but a fitness defect at OD600 = 2.4 (i.e. stationary phase). Genes in group 2 had negative fitness scores throughout the growth curve. These two groups contained the majority of mutants. Group 3 (four genes) had positive fitness scores in

61 log phase and a negative fitness score in stationary phase, while group 4 (three genes) had null or positive fitness scores at all phases of the growth curve. We clustered these genes by predicted functional category (Figure 6.15A, Data Set 2) and found that genes encoding purine metabolism enzymes and tRNA modification enzymes were enriched in group 1. However, the gene with the lowest fitness score in stationary phase, BOV_RS06060 (old locus tag BOV_1224), is annotated as serine-O-acetyltransferase (cysE) (Figure 6.14 and Figure 6.15B). As such, we chose to further characterize the function of cysE in B. ovis.

3.3.2. B. ovis ∆cysE enters stationary phase prematurely and has reduced culture yield in vitro

B. ovis CysE has high sequence identity (52%) and similarity (73%) with the well- characterized CysE enzymes of Escherichia coli and Salmonella enterica (248), and is clearly classified as CysE in the NCBI conserved domain database (E-value = 4.3e-122, https://www.ncbi.nlm.nih.gov/Structure/cdd). This protein is therefore predicted to execute the initial step in cysteine biosynthesis, specifically the addition of an acetyl group from acetyl-CoA to serine, producing O-acetylserine (Figure 6.16A). To confirm the cysE stationary phase phenotype observed in the RB TnSeq experiment (Figure 6.14), we built a B. ovis strain harboring an in-frame deletion of cysE (∆cysE). We grew ∆cysE in parallel with wild-type B. ovis ATCC

25840 (WT, Figure 6.16B) and observed that ∆cysE enters stationary phase earlier and terminates growth at a lower density than WT, thus corroborating the TnSeq result. This phenotype was complemented by the addition of 4 mM cysteine to the medium (Figure 6.16B) or by ectopic expression of the cysE gene from a lac promoter in the presence of IPTG (Figure 6.16C). Ectopic overexpression of cysE in a WT B. ovis background did not modify growth kinetics or the growth

62 curve shape compared to an empty vector control (WT/pSRK-EV) (Figure 6.16C). We conclude that cysE and cysteine biosynthesis are necessary for normal B. ovis growth yield in Brucella Broth.

3.3.3. cysK1 and cysK2 function redundantly in cysteine biosynthesis

CysK catalyzes the step in cysteine biosynthesis subsequent to CysE, namely the elimination reaction in which the acetyl group on O-acetylserine is displaced by sulfide to form cysteine (248) (Figure 6.16A). Given the stationary phase phenotype of ∆cysE, and the fact that this defect was chemically complemented by cysteine, we expected that mutations in cysteine synthase (cysK) should phenocopy ∆cysE. However, strains with transposon insertions in locus

BOV_RS09280 (old locus tag BOV_1893), annotated cysK in the NCBI RefSeq database, grew like wild type (Data Set 1). Growth of a strain harboring an in-frame deletion of BOV_RS09280 also grew the same as WT B. ovis in Brucella Broth (Figure 6.17A), confirming the TnSeq result.

We considered that the lack of an apparent growth defect in the ∆cysK strain may be due to the presence of other genes with CysK activity. A possible candidate for such a gene is locus

BOV_RS05050 (old locus tag BOV_1018), which encodes a protein with 37% sequence identity and 53% similarity to BOV_RS09280. Hereafter, we refer to BOV_RS09280 as cysK1 and

BOV_RS05050 as cysK2. Strains with transposon insertions in cysK2 also yielded a wild type phenotype in our TnSeq experiment (Data Set 1) and a strain harboring an in-frame deletion of cysK2 (∆cysK2) likewise grew the same as WT B. ovis. However, a ∆cysK1 ∆cysK2 double deletion strain exhibited a stationary phase/growth yield phenotype similar to ∆cysE (Figure

6.17A). Like ∆cysE, the ∆cysK1 ∆cysK2 phenotype was chemically complemented by the addition of cysteine to the medium (Figure 6.17B). The growth defect of the ∆cysK1 ∆cysK2 strain was genetically complemented by expressing either cysK1 or cysK2 from a plasmid (Figure 6.17C).

We conclude that these two genes have redundant function in the cysteine biosynthesis pathway.

3.3.4. B. ovis ∆cysE and ∆cysK1 ∆cysK2 strains are sensitive to exogenous H2O2 stress

Cysteine is, of course, important for protein synthesis. It is also one of the three amino acids that comprise glutathione (GSH) (Figure 6.18A), which plays a central role in the mitigation of a variety of stressors in bacteria (124) including oxidative stress. We hypothesized that defects in cysteine biosynthesis would have consequences on GSH synthesis and sensitize cells to oxidative stress. We thus attempted to complement the ∆cysE growth phenotype by adding GSH to the medium (Figure 6.18B). GSH supplementation partially complemented the ∆cysE growth yield defect. GSH limitation may directly contribute to premature entry of B. ovis ∆cysE into stationary phase, or GSH addition may restore cysteine homeostasis upon GSH catabolism. Since

GSH is known to be involved in decomposition of hydrogen peroxide to water (124) (Figure

6.18A), we assessed whether ∆cysE was more sensitive to H2O2 stress. We grew WT and ∆cysE to stationary phase, washed the cells, treated them with H2O2 for one hour in phosphate buffered saline solution, and then enumerated CFU. The ∆cysE strain was »2000 times more sensitive to

H2O2 than WT, and this sensitivity was rescued by either the addition of cysteine or glutathione to the medium during growth (Figure 6.18C). The hydrogen peroxide sensitivity phenotype of ∆cysE was genetically complemented by expression of cysE from the lac promoter of the pSRK plasmid

(∆cysE/pSRK-cysE) (Figure 6.18D). We further tested the sensitivity of the ∆cysK1 ∆cysK2 double deletion mutant to hydrogen peroxide treatment. Like ∆cysE, the ∆cysK1 ∆cysK2 strain was highly sensitive to peroxide treatment. The ∆cysK1 ∆cysK2 peroxide survival phenotype was chemically complemented by the addition of cysteine or glutathione to the medium (Figure

6.18E), and was genetically complemented by ectopic expression of either cysK1 or cysK2 (Figure

6.18F).

3.3.5. ∆cysE and ∆cysK1 ∆cysK2 have reduced viability in the intracellular niche

Brucella spp. primarily reside inside mammalian host cells. There are many challenges to growth and survival in the intracellular niche including nutrient limitation and exposure to stressors such as reactive oxygen species (ROS) (101). Given the in vitro growth and hydrogen peroxide sensitivity phenotypes of the cys mutants, we tested whether fitness of the ∆cysE and ∆cysK1

∆cysK2 strains was compromised in the intracellular niche. We infected a human monocytic cell line, THP-1, that we differentiated into macrophage-like cells. Although entry (2 hrs post- infection, p.i.) to the macrophage was unaffected by deletion of cysE or cysK1 and cysK2, there was a significant loss in recoverable colony forming units (CFU) of the cys mutant strains by 24 hrs p.i. relative to WT (Figure 6.19A and B). Thus B. ovis ∆cysE and ∆cysK1 ∆cysK2 enter host cells like WT but their fitness is compromised after entry.

In an effort to distinguish the relative contributions of intracellular killing and cysteine

(nutritional) limitation on reduced fitness of the cys mutants, we enumerated CFU recovered from

THP-1 cells at timepoints between 2 and 24 hrs p.i. Wild type and ∆cysE exhibit identical CFU loss between 2 and 8 hrs post-infection. WT begins to replicate by 12 hrs, but ∆cysE CFUs continue to decrease up to 24 hrs (Figure 6.19C). The rate of recoverable CFU increase between 24 hrs and

48 hrs p.i. is similar between WT and the cys mutants providing evidence that B. ovis has access to cysteine in this environment (Figure 6.19A and B). An attempt to interrogate a later time point

(72 hrs) was confounded by egressing bacteria and subsequent killing by gentamicin in the tissue culture medium (Figure 6.20). The intracellular infection defect we observe was complemented by expression of cysE from a plasmid (Figure 6.21A). We attribute partial complementation to the fact that cysE was expressed from a lac promoter on a replicating plasmid; there are challenges with full induction of transgenes from heterologous promoters in an intracellular infection context.

Given the ability of Brucella to infect multiple mammalian cell types, we next tested whether the in vitro infection phenotype of the cys mutants was particular to macrophages. We infected a sheep testis epithelial cell line (OA3.ts) (249), which is derived from a host tissue type that is relevant to B. ovis infection. OA3.ts entry was unaffected by the lack of cysE, but recovered

CFU were significantly lower for ∆cysE by 24 hr p.i. (Figure 6.19D). This phenotype was partially complemented by ectopic expression of cysE from a plasmid (Figure 6.21B). The magnitude of the ∆cysE defect 24 hrs p.i. was greater in THP-1 macrophages than in the OA3.ts epithelial line

(about 64-fold vs 16-fold, respectively; Figure 6.22). These in vitro infection data provide evidence that an intact cysteine metabolism system promotes B. ovis fitness in intracellular niche of multiple mammalian cell types.

3.4. DISCUSSION

3.4.1. A genome-scale search for B. ovis stationary phase mutants leads to cysteine metabolism

B. ovis is a widespread ovine pathogen that remains an understudied member of the

Brucella genus. Using a RB TnSeq approach, we sought to identify genes that are important for B. ovis growth and/or survival in the late phase of axenic broth culture (i.e. stationary phase), with a larger goal of uncovering genes that are important for fitness in the intracellular environment. We identified multiple genes for which Tn-himar disruption resulted in reduced fitness in stationary phase. Among the expected mutants in this dataset is rsh (BOV_RS03230), which controls the stringent response (25). Additionally, genes involved in purine metabolism, including purF have reduced fitness in dense culture. In Mycobacterium smegmatis, PurF influences survival during stationary phase (250), and purine metabolism is known to be important for growth of multiple

66 microbes in the intracellular and extracellular environments (251, 252). Multiple genes with a predicted role in tRNA modification also had diminished fitness in stationary phase. Transfer RNA modification enzymes have roles in translation quality control and can function to direct translation of specific transcripts under particular growth conditions (253). Given the phenotypes of tRNA modification mutants in stationary phase, it may be the case that these genes play a role of regulation of Brucella ovis physiology in the intracellular niche.

Tn-himar strains with insertions in cysE had the most diminished fitness in stationary phase, and cysE was therefore selected for follow-up studies. Sulfur and cysteine metabolism are central to microbial growth, and have been well studied in numerous pathogens (138). In Brucella spp., our understanding of sulfur metabolism is relatively limited though biosynthesis of sulfur- containing amino acids - cysteine and methionine - has been implicated in Brucella melitensis 16M infection of mice (254). Based on the high level of sequence identity/similarity to well-studied

CysE enzymes and established structural data on B. abortus CysE (255), B. ovis CysE is presumed to catalyze biosynthesis of O-acetylserine from acetyl-CoA and serine. The subsequent step in biosynthesis of cysteine from O-acetylserine requires displacement of the acetyl group by sulfide, a reaction that is catalyzed by CysK in many bacteria. Our growth data clearly implicate cysE in the cysteine biosynthesis pathway, as the in vitro growth defect of ∆cysE is rescued by the addition of cysteine. These results support published data that Brucella spp. can assimilate cysteine as an exogenous organic sulfur source (244, 256).

Surprisingly, the growth phenotypes of strains with Tn-himar insertions in the gene annotated as cysK in the RefSeq database did not differ from wild type, which suggested redundancy at this biosynthetic step. Consistent with this hypothesis, we have presented genetic evidence that two related enzymes, CysK1 and CysK2, function redundantly to produce cysteine.

It is possible that CysK1 and CysK2 do not catalyze the same reaction, but rather determine cysteine biosynthesis through two distinct routes. A recent report of such a case is the cystathionine b-synthase of Helicobacter pylori, which retains some O-acetylserine sulfhydrylase activity (257); this enzyme shares primary structure features with B. ovis CysK2. CysK-family enzymes can also have functions beyond direct involvement in cysteine metabolism (258), which may influence interpretation of our results. Notably, B. abortus CysE (serine O-acetyltransferase) and CysK2 do not form a cysteine synthase complex (CSC) in vitro (259). This supports a model in which CysK2 participates in cysteine synthesis via a mechanism that differs from that catalyzed by the typical

CysE-CysK CSC. We postulate that CysK1, rather than CysK2, binds to CysE to form the CSC in

Brucella. The development of a defined medium that supports the growth of B. ovis would greatly facilitate future study of CysK1 and CysK2 functions in cells. Exploration of the possible intracellular fitness advantage gained by redundancy at the CysK step of cysteine biosynthesis is an interesting area of future investigation.

3.4.2. Cysteine, glutathione and hydrogen peroxide stress

The growth and peroxide survival defects of ∆cysE were partially rescued by addition of cysteine or glutathione to the medium. Though elevated intracellular cysteine enhances susceptibility to hydrogen peroxide stress in Escherichia coli (126, 260), we do not observe peroxide sensitization of WT or ∆cysE B. ovis upon addition of 4 mM cysteine. 4 mM cysteine was consistently more protective than 4 mM GSH in our assay. GSH is an important redox control molecule, but the protective effect of GSH supplementation against H2O2 may be indirect.

Specifically, it’s possible that B. ovis transports and metabolizes some of the GSH to release cysteine, which is one of the three component amino acids of GSH. B. ovis is predicted to encode a g-glutamylcyclotransferase (BOV_RS09395), which catalyzes the cleavage of GSH to form

68 pyroglutamic acid and L-cysteinylglycine (123). The L-cysteinylglycine dipeptide could then be separated by peptidases to release cysteine. We nonetheless favor a model in which diminished

GSH production (as a result of abolished cysteine production) in ∆cysE directly affects H2O2 detoxification and growth yield of ∆cysE. Glutathione metabolism is important in B. ovis: the gshA biosynthesis gene is essential based on our previously published TnSeq dataset (161). Moreover,

Tn-himar insertions in BOV_RS04850 (old locus tag, BOV_0978), which is predicted to encode a glutathione-disulfide reductase (gor) – that reduces GSSG to GSH – resulted in a significant fitness disadvantage throughout the growth curve .

3.4.3. Cysteine and growth and the intracellular niche

Our study provides evidence that cysteine biosynthesis contributes to B. ovis fitness inside mammalian host cells. Strains harboring deletions of cysE or both cysK1 and cysK2 were not defective in host cell entry but had significantly reduced recoverable CFUs at 24 and 48 hrs post infection in human macrophage-like and ovine epithelial cell lines. Reduced recoverable CFU of the cys mutants at 24 and 48 hours can be attributed to defects that are manifested between 2 and

24 hours post-infection. It is difficult to fully discern the relative contributions of nutritional restriction and enhanced oxidative stress sensitivity to attenuation of ∆cysE and ∆cysK1 ∆cysK2 in vitro. The similar observed rate at which recoverable CFUs increase between 24 and 48 hours suggests that cysteine levels are not limiting – at least after 24 hours. The ER-derived replicative

Brucella containing vacuole (rBCV) supports bacterial replication (96) and can be established as early as 12 hrs p.i. (97); based on our data, we conclude that this compartment contains enough cysteine or cysteine-containing peptides to support growth. Of note, the defect of the ∆cysE strain was more pronounced in macrophage-like cells than in an ovine testis epithelial line. Sensitivity of ∆cysE and ∆cysK1 ∆cysK2 to ROS may underlie this difference in fitness between cell lines as

69 macrophages typically have a more robust respiratory burst than epithelial cells (261). A significant decrease in recoverable CFUs in ∆cysE is evident by 4 hr post-infection. This is a timepoint before intracellular Brucella replication occurs (97), supporting a model in which increased sensitivity to host killing underlies the reduced fitness of ∆cysE and ∆cysK1 ∆cysK2 in host cells.

Previous B. abortus TnSeq studies by Sternon and colleagues (262) did not identify cysE as a gene that was important for infection of Raw 264.7 macrophages. We observed significantly reduced B. ovis ∆cysE CFU relative to WT in Raw 264.7 cells by 24 hrs (Figure 6.23), but the

Sternon et al. experiment and our experiment differ in several ways. Indeed, the importance of cysteine metabolism in intracellular growth and/or survival may vary between B. ovis and B. abortus and between mammalian cell lines. Nonetheless, our data clearly provide evidence that a cysteine anabolism pathway in B. ovis is important for growth, stress survival and fitness in the intracellular niche.

Cysteine and methionine metabolic pathways are attractive targets to combat various pathogens (138) because, in mammals, these amino acids must be acquired from diet. Thus, compounds that disrupt cysteine metabolism are not predicted to have direct negative effects on mammalian metabolism. In fact, O-acetylserine sulfhydrylase (OASS; i.e. cysK) inhibitors are under investigation as therapeutics for Mycobacterium tuberculosis infections (140). Our work shows that genetic disruption of cysteine biosynthesis leads to a significant defect in B. ovis fitness within host cells. This pathway is therefore a possible target for combating brucellosis.

3.5. MATERIALS AND METHODS

Bacterial strains and growth conditions

Brucella ovis was grown on Tryptic Soy Agar (TSA, Difco Laboratories) plates, supplemented with 5% sheep blood (Quad Five) or in Brucella Broth (BB, Difco Laboratories, dissolved in tap water) for liquid cultures. Cells were incubated at 37 °C with 5% CO2 supplementation. Kanamycin (Kan) 50 µg/ml, sucrose (5% w/v) or isopropyl b-D-1- thiogalactopyranoside (IPTG, GoldBio) at 1 mM or 2 mM, were added when required.

Escherichia coli strains were grown in lysogeny broth (LB, Fisher Bioreagents) or on LB + 1.5% agar (Fisher Bioreagents) plates at 37 °C with Kan supplemented at a concentration of 50 µg/ml when required. E. coli WM3064 strain, used for conjugation, was grown in the presence of 300

µM diaminopimelic acid (DAP, Sigma-Aldrich), as it is a DAP auxotroph.

Plasmid and strain construction

Deletion plasmid construction

To build the deletion strains, fragments of approximately 500 bp upstream and downstream of target genes were amplified with KOD Xtreme Hot Start polymerase (Novagen). These fragments were built so that 9 bases at both the 5’ and 3’ ends of the gene were maintained, keeping the gene product in frame to minimize polar effects. Purified DNA from Brucella ovis ATCC

25840 was used as a template. Amplified fragments were gel purified (ThermoFisher Scientific) and assembled into the pNPTS138 suicide deletion vector (digested with HindIII and BamHI restriction enzymes, New England Biolabs) using Gibson assembly (New England Biolabs).

Complementation plasmid construction

To build plasmids for genetic complementation, cysE, cysK1 or cysK2 were PCR amplified from B. ovis ATCC 25840 with KOD Xtreme Hot Start polymerase, gel purified (ThermoFisher

Scientific) and Gibson assembled into pSRK (263) that had been digested with NdeI and KpnI

71 restriction enzymes (New England Biolabs). cysE, cysK1 or cysK2 were cloned downstream of Plac

(lactose, IPTG inducible promoter).

Delivery of plasmids to B. ovis

Constructed plasmids were transformed into chemically competent E. coli Top10 strains for plasmid maintenance. All plasmid inserts were confirmed by PCR and Sanger sequencing, and plasmids were delivered to B. ovis by conjugation using E. coli WM3064 as a donor strain. For conjugation, WM3064 donor strains were mated with B. ovis strains and spotted on TSA blood plates plus DAP and incubated overnight at 37 °C in a 5% CO2 atmosphere. Mating spots were spread on TSA blood plates plus Kan (without DAP) to select for B. ovis plasmid acquisition.

When deleting genes using the pNPTS138 plasmid, merodiploid clones were inoculated in

Brucella Broth overnight to allow for a second crossover event, then spread on TSA blood plates plus sucrose (5% w/v) for counterselection. Single colonies harboring gene deletions were identified by patching clones on TSA blood plates with or without Kan. The putative deleted locus was PCR amplified using gene-flanking primers in Kan-sensitive clones, and the PCR fragment was resolved by gel electrophoresis to test whether the gene had been deleted. For a complete list of strains, plasmids, and primers, please see Data Set 3.

Growth Curves

Cells were inoculated from »48hr-old TSA blood plates into BB at densities ranging from

OD600 0.08 to OD600 of 0.2. Growth was assessed spectrophotometrically measuring optical density at 600 nm (OD600). Growth curves were conducted at least three independent times with two or three technical replicated in each experiment. Representative curves are shown for each set of

72 strains. Where indicated, cysteine (4 mM), GSH (4 mM), Kan (50 µg/ml), or IPTG (1 mM) were supplemented upon start of growth experiment to the liquid media.

H2O2 survival assays

Cells were grown overnight in BB to stationary phase (OD600 of »2). Cells were pelleted and resuspended in Phosphate-Buffered Saline (PBS, Sigma) to achieve an OD600 of 0.15. 200 µl of cells were added to 1.8 ml of PBS or PBS supplemented with fresh H2O2 (15 or 20 mM final concentration), bringing the final OD600 to 0.015. Cells were then incubated at 37 °C in 5% CO2 for 1 hr before spotting aliquots of a 10-fold serial dilution series on TSA blood plates. CFUs were enumerated after 48 hrs incubation at 37 °C in 5% CO2. Experiments were repeated at least three times with each sample in duplicate or triplicate in each experiment.

DNA extractions

Cells from 1 ml of stationary phase culture were pelleted by centrifugation, washed once in PBS, and resuspended in 100 µl TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) supplemented with 1 µg/ml RNAseA. Cells were lysed by addition of 0.5 ml GES lysis solution

(5 M guanidinium thiocyanate, 0.5 M EDTA pH 8.0, 0.5% v/v Sarkosyl) and 15 min incubation at

60 °C. 0.25 ml cold 7.5 M ammonium acetate (Fisher Bioreagents) was added, and mixture was incubated on ice for 10 min. 0.5 ml of chloroform (Fisher Bioreagents) was added to separate the

DNA, samples were vortexed and centrifuged. Aqueous top phase was moved to a fresh 1.5 ml centrifuge tube and 0.54 volumes of cold isopropanol was added to precipitate the DNA. After centrifugation, isopropanol was discarded and pellets were washed three times in 70% ethanol before resuspending pellets in TE buffer. Concentration and purity of the extracted DNA were determined spectrophotometrically (NanoDrop One, Thermo Scientific).

Barcoded TnSeq

A B. ovis RB Tn-himar library was built and mapped as described (244). Briefly, E. coli

APA752 (a WM3064 donor strain carrying a pKMW3 mariner transposon library) was conjugated into B. ovis bcaA1 (161) under atmospheric CO2 conditions. Kan resistant colonies were collected, grown to OD600 = 0.6 and frozen in 1 ml aliquots. Genomic DNA was extracted and the Tn insertion sites mapped as previously described (243).

To identify genes that confer a fitness advantage in stationary phase, the B. ovis Tn-himar library was inoculated in BB in a 5% CO2 environment in triplicate at an OD600 of 0.0025. An aliquot of each initial culture was collected as the reference time point. Cultures were then grown to stationary phase, with samples harvested throughout the growth curve at OD600 of 0.05, 0.12,

0.9 and 2.4. Cells from each sample were pelleted by centrifugation and resuspended in water.

Barcodes were amplified from approximately 1.5 ´ 108 cells per PCR reaction (see Table 7.6) with primers that both amplified the barcodes and added indexed adaptors (243). Amplified barcodes were then pooled, purified and sequenced on an Illumina HiSeq 4000.

Fitness scores for each gene were calculated following the protocol of Wetmore and colleagues (243) using scripts available at https://bitbucket.org/berkeleylab/feba. Genes for which

2 out of 3 samples had at least one timepoint with a t-like statistical significance score (243) ³ |4| were included in subsequent analyses. A heat map of fitness scores of genes passing this filter is shown in Figure 6.13 and the raw fitness data are in Data Set 1.

Finally, we averaged the fitness values of the three replicates and kept mutants that had an average fitness score ³ |1| in at least one time point. Mutants in this group with a standard deviation

³0.75, were manually inspected and extreme outlier points were removed from a total of six genes.

The genes with adjusted average and standard deviation values are shown in red in Data Set 1.

Heat map of averaged fitness values is shown in Figure 6.14.

Tissue culture

All tissue culture cells were grown at 37 °C with 5% CO2 supplementation. THP-1 cells

(ATCC TIB-202) were cultured in Roswell Park Memoriam Institute medium (RPMI 1640, Gibco)

+ 10% Fetal Bovine Serum (FBS, Fisher Scientific). The RAW 264.6 (ATCC TIB-71) and the

OA3.ts (ATCC CRL-6546) cells were grown in Dulbecco’s Modified Eagle Medium (DMEM,

Gibco) supplemented with 10% FBS.

Infection assays

THP-1 cells were seeded at a concentration of 105 cells/well in 96-well plates and phorbol myristate acetate (PMA) at final concentration of 50 ng/µl was added to induce differentiation into macrophage-like cells for 48-96 hrs prior to infection. OA3.ts cells were seeded at a density of 5

´ 104 cells/well in 96-well plates for 24 hrs prior to infection. B. ovis cells were resuspended from a 48 hr old plate in RPMI + 10% FBS or DMEM + 10% FBS and added to tissue culture plates on the day of infection at multiplicity of infection (MOI) of 100 for THP-1, and at an MOI of 1000 for OA3.ts cells. When infecting with the complementation strains carrying the pSRK plasmid, the

Brucella strains were struck on TSA blood plates ith Kan and IPTG 48 hrs prior to infection, and

2 mM IPTG was added to the tissue culture media throughout the duration of the experiment.

Plates were spun for 5 min at 150 ´g and incubated for 1 hr at 37 °C in 5% CO2. Fresh media was supplied containing 50 µg/ml of gentamicin and incubated for another hour. Cells were then washed once with PBS and once in H2O and then lysed with H2O for 10 min RT at 2 hrs, 24 hrs, and 48 hrs post infection. Lysates were serially diluted, spotted on TSA blood plates and incubated

75 at 37 °C in 5% CO2 for 48 hrs to enumerate CFUs. Experiments were repeated at least 3 times with three technical replicates.

3.6. SUMMARY

Brucella ovis is an ovine intracellular pathogen with tropism for the male genital tract. To establish and maintain infection, B. ovis must survive stressful conditions inside host cells, including low pH, nutrient limitation, and reactive oxygen species. These same conditions are often encountered in axenic cultures during stationary phase. Studies of stationary phase may thus inform understanding of Brucella infection biology, yet the genes and pathways that are important in Brucella stationary phase physiology remain poorly defined. We measured fitness of a barcoded pool of B. ovis Tn-himar mutants as a function of growth phase and identified cysE as a determinant of fitness in stationary phase. CysE catalyzes the first step in cysteine biosynthesis from serine, and we provide genetic evidence that two related enzymes, CysK1 and CysK2, function redundantly to catalyze cysteine synthesis at steps downstream of CysE. Deleting either cysE (∆cysE) or both cysK1 and cysK2 (∆cysK1 ∆cysK2) results in premature entry into stationary phase, reduced culture yield and sensitivity to exogenous hydrogen peroxide. These phenotypes can be chemically complemented by cysteine or glutathione. ∆cysE and ∆cysK1 ∆cysK2 strains have no defect in host cell entry in vitro but have significantly diminished intracellular fitness between 2 and 24 hours post infection. Our study has uncovered unexpected redundancy at the

CysK step of cysteine biosynthesis in B. ovis, and demonstrates that cysteine anabolism is a determinant of peroxide stress survival and fitness in the intracellular niche.

4. DISCUSSION, CHALLENGES FOR THE FUTURE, AND

CONCLUSIONS

Our work in Brucella ovis in the context of different environmental stimuli and its capacity to interact and react to these factors has increased our understanding of this furtive bacterium. In

Chapters II and III I detailed our discoveries in relation to CO2 sensitivity and the implications of cysteine metabolism within the host intracellular environment. In this final chapter, I will briefly discuss some challenges and possible future directions.

4.1. CO2 SENSITIVITY

In Chapter 2, I addressed a historical observation about CO2-dependence of different

Brucella species and biovars. In trying to understand the genetic determinants that underlie B. ovis sensitivity to CO2 levels, we uncovered a complex collection of alleles within different Brucella lineages. We show that these alleles determine Brucella capacity to grow in atmospheric conditions. Furthermore, when the allele in question encodes for a non-functional carbonic anhydrase, this sensitizes the strain to CO2 levels, with implications at the transcriptional level for a multitude of genes, including those affecting virulence.

The increased sensitivity in Brucella ovis to partial CO2 pressures is an interesting discovery. Indeed, the bcaABOV pseudogene may be under neutral selection and in the process of being lost within the Brucella population, with Brucella ovis and most B. abortus biovars as the forerunners of this process. There is no evident morphological difference between WT B. ovis and

B. ovis bcaA1 strains grown in 5% CO2 (Figure 6.24), as well as no difference at the transcriptional

77 level (see Figure 6.8 and Figure 6.9). The lack of an obvious in vivo advantage (162) to harboring the non-functioning carbonic anhydrase pseudogene provides evidence that this model is accurate.

Of note, García-Lobo and colleagues did discern a slight fitness advantage for CO2-dependent

Brucella abortus 292 strain versus a derived CO2-independent mutant (264). This experiment was conducted in female BALB/c mice, where spleen and lungs were tested for Brucella colonization by CFU plating after 8 weeks of infection. To distinguish between the two strains, dilution plates were either grown in the presence or absence of CO2 supplementation. While these results are interesting and offer a potential in vivo advantage to harboring the non-functional bcaA pseudogene allele, it should be noted that there are a couple of points to this experiment that should be addressed and perhaps more fully explored. On one hand, neutral selection may be causing an increase of the ratio of CO2-dependent bacteria (from the CO2-independent population) once the inocula is within the host (and in a high CO2 environment), so perhaps using a different selection method to distinguish between the two inoculated strains may be more apt, like different antibiotic resistance genes stably transformed within the strains. Furthermore, the bcaA pseudogene (and consequently CO2-dependence) is found in all B. ovis isolates (to date) and in most B. abortus biovars, but there is no evidence of other Brucella species harboring this non-functional allele. So, if indeed there is an advantage to losing carbonic anhydrase function, it might be linked to these two Brucella species. Since Brucella ovis and Brucella abortus do not share preferred host species,

LPS structure, nor are the closest kin based on 16S phylogenetic analysis, it is unlikely that a murine infection model will serve well as a basis to understand the cause and physiology of this loss-of-function selection process. Indeed, it is possible that a set of genes or gene functions that are shared between B. abortus and B. ovis, or some B. abortus biovars, determine an advantage to losing carbonic anhydrase function. If this were the case, is there residual function or a novel

78 function for the bcaA pseudogene? Again, that may be the case if the murine competitive infection assays are telling of what is happening in the environment during naturally-occurring Brucella infections. This might indeed be the reason we do not see more mutations accumulate within bcaABOV (as shown in Figure 6.12). Further studies, for instance at a population level, where freshly isolated strains were immediately screened for CO2 dependence and their bcaA alleles sequenced, might help understand this phenomenon better. Furthermore, the site of isolation might also be telling: for instance, whether B. abortus isolated from an aborted fetus carry more CO2- independent strains, as the Brucella now are in an environment with only 0.04% CO2, versus

Brucella abortus isolated from a lymph node or the host testes.

4.2. OTHER BRUCELLA CARBONIC ANHYDRASES

Interestingly, Brucella do seem to encode for at least three carbonic anhydrases, two members of the b-family (BcaA and BcaB) and a putative g-carbonic anhydrase (RicA) (209, 265,

266). RicA has never shown any carbonic anhydrase activity, and is one of the few identified effector proteins secreted by the T4SS (see Chapter 1, The intracellular niche). BcaA from B. suis 1330 has been biochemically characterized by Joseph et al. (214), and has modest CO2

6 -1 hydration activity, with a kcat of 1.1 ´ 10 s . I built a model structure for BcaA1BOV (Figure

6.25A) based on sequence homology to Pisum sativum (Figure 6.25B and C) with which it shares

~35% amino acid identity. The active site with the zinc bound to two cysteines and a histidine, along with the aspartic acid that acts as the fourth zinc ligand instead of a water molecule are shown in green in detail in Figure 6.25A right panel (characteristic of the type II b-carbonic anhydrase family, see Chapter 1 and also orange asterisks in alignment in Figure 6.4B). E. coli carbonic anhydrase, CynT, instead, shares 41% identity with the crystal structure of a lyase of

Synechocystis sp. (strain PCC 6803 / Kazusa), and is shown in Figure 6.25B and D), and shares

33% identity to Pisum sativum. The portions in teal for the three structures indicate the area that is affected by the frameshift in B. ovis. The crystal structure of the second E. coli carbonic anhydrase, Can, is also shown (Figure 6.25E). Based on the predicted structure (see also García-

Lobo et al. (264)) of BcaA1BOV, it seems plausible that the SNPs that transform BcaA from a functional carbonic anhydrase to a pseudogene in Brucella ovis do not affect the active site directly

(Figure 6.25, green), but rather affect dimerization (Figure 6.25, teal ribbons). While these mutations do, overall, affect function of the resulting enzyme, it may be that the bcaA pseudogene is still capable of binding CO2 or bicarbonate, which might in turn serve an ulterior function, such as sensing or sequestering these molecules. Further biochemical studies on this enzyme and crystallographic data might help better understand the nature of this protein. Of note, our attempts of purifying BcaABOV were unsuccessful, as it was mostly recovered in the insoluble fraction

(inclusion bodies, I.B.), and would rapidly precipitate out of solution (Figure 6.26 and data not shown). This was not the case for B. ovis BcaA1 or the Brucella suis 1330 carbonic anhydrase

BcaABSU, which was used as a positive control, as previous reports indicate successful purification

(207, 214). This could suggest that BcaABOV is indeed unable to dimerize, which is important for carbonic anhydrase function and enzyme stability. Finally, since carbonic anhydrases can also catalyze other hydration reactions other than CO2 hydration (see Chapter 1), it is possible that the bcaA pseudogene retains (or gains) this additional function while losing carbonic anhydrase activity. Further biochemical studies would be necessary to fully understand what the frameshift mutation that inactivates BcaA1BOV does to this enzyme.

The other Brucella b-carbonic anhydrase, bcaB, was also purified in B. suis 1330 by Joseph and colleagues (207). It showed a more modest activity compared to B. suis 1330 bcaA, with a kcat

80 of 6.4 ´ 105 s-1. Interestingly, the B. ovis bcaB allele is missing 24 nucleotides in the middle of the gene (starting at codon 70) compared to other Brucella alleles (Figure 6.27A). There are two amino acid substitutions between BcaB in Brucella suis and Brucella abortus that seem to affect activity (162). In fact, B. abortus BcaB (BcaBBAB), carries a glycine in position 76 instead of a valine (which is found in B. suis 1330 BcaB, BcaBBSU). Perez-Etayo and colleagues have shown that two different B. suis strains (which are CO2-independent) can still grow when lacking either b-carbonic anhydrase, that is BcaABSU or BcaBBSU. On the contrary, B. abortus CO2-independent strain, B. abortus 2308W, cannot grow in atmospheric conditions if bcaABAB2308W is deleted. The only common amino acid in the two B. suis BcaB alleles that is different in B. abortus is indeed the valine in position 76 (glycine in B. abortus), suggesting that the inactivity (or reduced activity) of BcaBBAB2308W is due to this amino acid substitution (162). Interestingly, this codon is not present in BcaBBOV, as it is in the middle of the 8 amino acid-deleted portion (Figure 6.27A), indicating the importance of this locus for this carbonic anhydrase. We expressed bcaBBAB2308 ectopically in

WT B. ovis to attempt to rescue growth when CO2 was not supplemented, but either bcaBBAB2308 is not functional in B. ovis or it is not active enough to compensate for the bcaA pseudogene, because its expression upon IPTG induction did not allow B. ovis to grow in atmospheric CO2 conditions (Figure 6.27B and C). Of note, Perez-Etayo et al. also found that bcaBBSU1330 did not rescue growth of wild type CO2-dependent B. abortus strains, which carried the bcaABAB pseudogene (see Figure 6.7). This may suggest either that BcaBBSU activity is not sufficient to rescue growth in B. abortus when the bcaA allele is non-functional or that the functional and non- functional monomers are interacting in a way that either affects dimerization or yields a non- functional multimer, as this experiment was not conducted in a ∆bcaA background (162). As of now, the exact role for bcaB is in CO2 fixation in these different Brucella species is not understood,

81 but further functional and structural characterization of this enzyme might lead to better understanding of the CO2 acquisition network in Brucella, and shed light on the species-specific behaviors.

4.3. EFFECTS OF ALTERED SULFUR AND CYSTEINE METABOLISM IN B. OVIS

In Chapter 3, we found that the disruption of cysE strongly affected stationary phase in

Brucella ovis. As cysE encodes for the enzyme responsible for the first committed step in de novo cysteine biosynthesis, this led us to investigate the role of sulfur metabolism in Brucella ovis.

Specifically, we noted redundancy in the pathway at the level of the cysteine synthase step (CysK) and assessed how resistance to hydrogen peroxide is affected by interfering with cysteine biosynthesis. Furthermore, lack of cysteine production hinders the capacity of B. ovis to survive and replicate the hostile intracellular environment.

Sulfur is essential for life and it is therefore not immediately surprising that disruption of sulfur metabolism (or metabolic pathways that rely on sulfur availability, such as cysteine biosynthesis) would lead to important detrimental effects within Brucella ovis. What was surprising was that an important gene involved in de novo cysteine synthesis from serine (cysE) emerged as the costliest loss in stationary phase. Cysteine biosynthesis pathways interconnect in various ways in different organisms, as not all enzymes involved in different metabolic pathways are always present. Therefore, the specific effects that come with alteration of cysteine metabolism in B. ovis will have B. ovis-specific consequences, which helped us understand in more detail the nature of this pathogen’s interaction with the challenging intracellular environment it encounters.

4.3.1. Coloration of ∆cysE strains

When scraping ∆cysE off of blood plates, we noticed an interesting occasional phenomenon: the pellet had a dark-purplish tint. Upon further attempts to investigate the phenomenon, we scraped pellets from different genetic backgrounds and compared the colors

(Figure 6.28A and B). Interestingly, it seemed that the dark, purplish color matched strains in which cysE was not expressed. In fact, either WT B. ovis or B. ovis/pSRK-EV, B. ovis

∆cysE/pSRK-EV, and B. ovis ∆cysE/pSRK-cysE (in the absence of IPTG) always had a paler-to- white coloring, compared to B. ovis ∆cysE or the complementation strain grown on IPTG plates.

Unfortunately, attempt to quantitatively describe this phenomenon failed, partially due to the transient nature of the color in the first place – cells would bleach rather quickly – and partially because of the difficulty in stably reproducing the phenomenon. We assessed that the time since thawing mattered, the chance of color being observed higher from strains struck 48 hours prior to collection, later times (when B. ovis colonies assume a more whitish coloring on the plates compared to the initial grayish hue) were never colored (Figure 6.28A, bottom). We did not determine a correlation with thickness of the lawn of Brucella (data not shown) nor with the presence of blood on the plate (Figure 6.28A). Nevertheless, it is an interesting phenotype that might be intriguing to pursue. Indeed, in the yeast pathogen Cryptococcus neoformans, Toh-e and colleagues were investigating sulfur amino acid metabolism and stumbled upon a similar observation: their ∆cysE or ∆cysK (cys2∆ and cys1∆, respectively), when struck on plates with complex media supplemented with 1 mM copper sulfate (CuSO4), acquired a rusty, dark-purple coloring, while wild-type C. neoformans remained white (267). This phenotype was not complemented by cysteine supplementation. The authors suggested the reason behind the coloring to revolve around H2S accumulation when cysteine biosynthesis was halted because O-acetylserine

83 was not available as an acceptor for the reduced sulfur. We did not detect changes in coloring specific to the use of plates supplemented with CuSO4 (data not shown).

4.3.2. Sensitivity to transition metals

In C. neoformans, it was also observed that interfering with cysteine biosynthesis affected sensitivity to heavy metals, such as copper (CuSO4) and cadmium (in the form of CdCl2). We first tested our ∆cysE strains with different concentrations of these metals by imbibing discs and analyzing the zone of clearance on a lawn of Brucella ovis (Figure 6.28C). We observed that in the presence of 10 mM cadmium chloride-imbued discs, there was a zone of clearing in B. ovis that was not clearly defined, while on lawns of B. ovis ∆cysE, a shiny metallic zone appeared

Figure 6.28C, white arrow), possibly suggesting metal precipitation. The zone of clearing round copper sulfate discs did appear slightly larger in strains lacking cysE compared to wild-type B. ovis (Figure 6.28C) and is in line with the observations made for Cryptococcus neoformans (267), though not as drastic. Of note, plates left in the incubator over time developed single defined colonies in the zone of inhibition of the CuSO4, albeit close to the perimeter, in both WT and ∆cysE strains, suggesting the birth of copper-resistant mutants (Figure 6.28C, white asterisks). The origin of this shiny metallic zone in the presence of cadmium is also an interesting area of future investigation.

4.4. CYSTEINE AND METHIONINE METABOLISM IN BRUCELLA OVIS

4.4.1. B. ovis presumably does not encode for genes required for the reverse transsulfurylation pathway

Cysteine may be synthesized via a second metabolic pathway, from methionine, known as the reverse transsulfurylation pathway. Specifically, this pathway involves are a cystathionine beta-synthase (CBS, or mccA), which produces cystathionine from homoserine (a methionine derivative) and serine, and a cystathionine g-lyase (CGL, or mccB), which yields cysteine (see

Chapter 1). To attempt to test whether the reverse transsulfurylation pathway is functional in B. ovis, we attempted to rescue the ∆cysE phenotype by supplementing methionine in the media.

Increasing amounts of methionine (up to 10 mM) did not restore growth of the cysE-null strain to

WT levels (Figure 6.29A). It is possible that the reason for this observation is that methionine is not imported within the cell. We deem this unlikely, as we identified a putative methionine transporter in the B. ovis genome (metQ, locus tag BOV_RS14915) and given the promiscuity of amino acid transporters in general (268). We also did not identify homologs to CBS and CGL in

B. ovis. Given these results, we assume that Brucella ovis does not have a functional reverse transsulfurylation pathway.

4.4.2. The redundancy of the two cysK genes in B. ovis

We found the functional redundancy of cysK1 and cysK2 in B. ovis with regards to cysteine biosynthesis intriguing. As mentioned in Chapter 1, cysteine biosynthesis can occur from serine

(de novo) or via the reverse transsulfurylation pathway (Figure 6.30). We did not find any genes that could putatively encode for enzymes involved in the reverse transsulfurylation pathway.

However, recent reports describe a novel peculiar cystathionine b-lyase which used OAS instead of serine to produce cysteine directly from homocysteine (O-acetylserine-dependent CBS, or

OCBS) (269–271). In 2019 same kind of enzyme was described in the enteric pathogen

Helicobacter pylori (257). Detailed analysis highlighted key amino acid portions that compare

OASS, CBS and OCBS. We wondered whether one of the two Brucella ovis genes identified as

85 showing cysK function (cysK1 and cysK2, see4Chapter 3), may actually encode for an OCBS.

Upon comparison of five significant regions the authors identified in H. pylori as determinants of the specific enzyme type and function, it seems that cysK1 looks more like an O-acetylserine sulfhydrylase (OASS), while cysK2 is more similar to an OCBS (

Figure 6.31 and ref). However, since B. ovis does not appear to encode for CGL, the role for cysK2 as an OCBS is obscure. More experiments would be necessary to understand the catalytic abilities of these important players, and a functioning minimal or determined media would aid future experiments greatly. To this end, testing homologous enzymes in B. abortus (for which a functional minimal media has been identified) might be a good starting point.

4.5. CONCLUSIONS

The role of sensitivity to environmental CO2 levels in B. ovis has shed further light in understanding the physiology and evolution of this outlier member of the Brucella family as well as providing a marker for select Brucella linages. The interconnection of sulfur metabolism and the capacity for Brucella ovis to survive its preferred environment is interesting to consider in the search for novel targets for antibiotics. It is also another window into the pathways that are involved in Brucella adaptation to hostile environments where it has evolved to thrive. Dissecting the nature of the response of Brucella to the intracellular niche, specifically the different stressors and growth phases, are increasing our understanding on how bacteria adjust to change and combat specific attacks.

My work has provided insight into the reactions to the different types of environments

Brucella ovis encounters during its life cycle. As a facultative intracellular bacterium, it is thoroughly challenged by the host cell environment and has thus evolved mechanisms that permit

86 it to thrive even in this hostile habitat. Sensing, persisting and reacting to its everchanging surroundings have enabled this furtive bacterium to establish a niche that ensures its survival and proliferation, granting it all the tools it requires to endure and evolve with, and perhaps beyond, its host.

5. APPENDIX A – REGULATION OF THE ERYTHROBACTER

LITORALIS DSM 8509 GENERAL STRESS RESPONSE BY

VISIBLE LIGHT

5.1. PREFACE

The contents of this appendix was adapted from its published form:

Fiebig A, Varesio LM, Alejandro Navarreto X and Crosson S. Molecular Microbiology (2019).

DOI:10.1111/mmi.14310

Additional supplementary information (Tables S1-S6) may be found online.

5.2. INTRODUCTION

5.2.1. LOV-HWE kinases: An overview

Light is a ubiquitous environmental factor that provides energy to support life in many ecosystems. Multiple adaptive responses to light have evolved in bacteria including phototaxis and photoavoidance, photoprotective pigment production, and the regulation of genes required for photosynthesis. These responses are mediated by proteins that sense photons in particular energy ranges across the visible spectrum (272). LOV domains are widely distributed photosensors (273) that detect blue light via a bound flavin cofactor (274). These photosensory domains are present in an assortment of signal transduction proteins from bacteria, archaea, fungi, protists, and plants

(275, 276). Though broadly conserved (277, 278), the physiological roles of LOV photosensors in bacteria remain largely undefined.

The class Alphaproteobacteria contains species that inhabit diverse ecological niches, and that have a significant impact on nutrient cycling, agricultural production and human health (279).

Alphaproteobacteria commonly encode proteins that contain a LOV domain coupled to an

HWE/HisKA2-type (280) histidine kinase. Surprisingly, these “LOV-HWE kinases” are often present in heterotrophic species with no evident photobiology (276). Histidine kinases are typically co-expressed with and phosphorylate their cognate response regulators to directly control gene expression (281), but HWE/HisKA2 kinases are unusual in that they are often orphaned on the bacterial chromosome, or are adjacent to single domain response regulators (SDRR) that lack a regulatory output domain to control gene expression (280).

Early studies of Alphaproteobacterial LOV-HWE kinases in Caulobacter crescentus (282) and Rhizobium leguminosarum (283), demonstrated their influence on cell adhesion and biofilm formation. In the case of C. crescentus, a LOV-HWE kinase (LovK) modulates adhesion by controlling expression of a single downstream gene that regulates surface adhesin biosynthesis

(284, 285). However, the major effect of deleting or overexpressing genes encoding LOV-HWE kinases appears to be dysregulation of the general stress response (GSR) system (246, 286), which determines cell survival across a range of stress conditions (40, 42). There is increasing evidence that multiple HWE/HisKA2-family kinases function as part of complex regulatory networks that regulate the GSR in Alphaproteobacteria by influencing the phosphorylation state of the anti-anti- s factor, PhyR (287–290). Phospho-PhyR activates an extracytoplasmic function (ECF) s factor

– EcfG – by binding and sequestering its anti-s factor, NepR (291) (Figure 6.32).

Though there is genetic evidence that LOV-HWE kinases control transcription of genes in the GSR regulon (246, 286, 287, 292), the effects of visible light on LOV-HWE kinase signaling in vivo remain undefined. Therefore, it is important to identify appropriate experimental systems

89 to assess the effect of visible light and other environmental signals on LOV-HWE kinase signaling in living cells.

5.2.2. Erythrobacter litoralis DSM 8509 as a system to study LOV-HWE kinase signaling

We sought to investigate the signaling role of LOV-HWE kinases in the genus

Erythrobacter, which was attractive to us for several reasons. Erythrobacter spp. are abundant in the world’s oceans (278, 293, 294) where they contribute to global nutrient and energy cycling

(295, 296). Importantly, LOV-HWE kinases are common in Erythrobacter spp., as evidenced in both sequenced isolates (297, 298) and in a set of metagenome assembled genomes (MAGs) (299) from the Tara oceans sequence datasets (278). Though Erythrobacter spp. can be isolated and grown in axenic culture (298, 300), strains have not been extensively cultivated and manipulated.

Thus, Erythrobacter spp. are less likely to have acquired lab-adaptive mutations that could potentially skew the effects of light on cell physiology. Most Erythrobacter spp. are aerobic anoxygenic photoheterotrophs (AAP), which carry a photosynthesis gene cluster that encodes production of bacteriochlorophyll a (Bchl a), and other genes required for phototrophy. Therefore, physiological responses to visible light are expected in this genus.

Erythrobacter litoralis strain DSM 8509 was isolated from a cyanobacterial mat in the supralitoral zone off the island of Texel, Netherlands (301, 302). Draft genome sequence of DSM

8509 indicated the presence of a single LOV-HWE kinase gene (298). This distinguishes DSM

8509 from E. litoralis strain HTCC 2594, which encodes multiple LOV-kinases (297) that have been previously characterized in vitro (290, 303, 304). HTCC 2594 does not encode the genes required for phototrophy (297, 305), and though DSM 8509 and HTCC 2594 cluster phylogenetically based on their 16S sequences, they do not group based on amino acid sequence

90 of concatenated core genes (305). As such, it has been suggested that HTCC 2594 should be re- classified as a distinct species, Erythrobacter sp. HTCC 2594 (306).

Here, we report the development of E. litoralis DSM 8509 as an experimental genetic system, which we have used to study the contribution of a LOV-HWE kinase to regulation of the general stress response. Our data show that an ensemble of three HWE kinases, including the LOV-

HWE kinase LovK, collectively regulates the general stress response in E. litoralis. The cytoplasmic kinase, GsrK, functions as an activator of GSR transcription in complex medium, while the transmembrane kinase, GsrP, is a repressor. LovK activates GSR transcription, and is a more potent GSR activator under dark conditions. Transcription of the entire GSR regulon is significantly higher in dark conditions than in light. In fact, the set of genes that are differentially transcribed upon shifts in the light environment strongly overlaps the experimentally-defined GSR regulon. While LovK contributes to light-dependent regulation of GSR transcription, it is not strictly required for this response. Our results support a model in which photons directly and indirectly modulate GSR transcription via the HWE kinases GsrK, GsrP, and LovK.

5.3. RESULTS

5.3.1. A brief comparison of Erythrobacter litoralis DSM 8509 to other Erythrobacter species

Several partial and complete genome sequences of Erythrobacter isolates have been deposited in public databases (see Table 7.7 for representative genomes). These genomes have similar characteristics in terms of size and GC content, yet differ in presence/absence of particular genes and pathways. For example, Erythrobacter spp. can encode zero, one, or multiple LOV-

HWE kinases, and the total number of HWE/HisKA2-family kinases ranges from two to sixteen

(307). Although the three LOV kinases present in strain HTCC 2594 have been biochemically

91 characterized (290, 303, 304), we anticipated that the potential functional redundancy of these genes could complicate interpretation of genetic and physiological studies in vivo. In contrast,

DSM 8509 encodes only one LOV-HWE kinase. Additionally, when considering our intent to investigate the interplay between multiple HWE/HisKA2-family kinases involved in GSR regulation, we noted that DSM 8509 has a ‘goldilocks’ number of such kinases: more than one, but not too many to complicate genetic analyses (Table 7.9). Finally, DSM 8509 encodes core genes required for phototrophy, which provides an opportunity to develop genetic tools in a species for which light is likely a central environmental/metabolic signal. For these reasons, we chose to pursue the development of E. litoralis DSM 8509 as a comparative model to study the function of

LOV-HWE kinases.

5.3.2. Whole-genome sequencing and development of genetic tools in DSM 8509

To our knowledge, genetic analysis of an Erythrobacter species has not been reported. To support development of Erythrobacter litoralis DSM 8509 as an experimental genetic model, we first completed and closed the genome sequence. Specifically, we re-sequenced DSM 8509 using a PacBio long-read sequencing platform to an average depth of 143X (Table S1). These sequence reads were assembled de novo to a single 3.25 Mb contig, which was deposited and is available to the public under NCBI GenBank accession CP017057. We surveyed the antibiotic sensitivity profile of DSM 8509 to identify useful markers for genetic selection (Table S2), and adapted methods to transform the bacterium with both replicating (pBBR-derived) and integrating (ColE1- derived) plasmids. We successfully transformed E. litoralis DSM 8509 by both electroporation and conjugation (see Materials and methods). We further identified conditions to generate unmarked gene deletion and allele replacement strains using a two-step recombination and

92 sacB/sucrose counterselection approach, which we applied to our genetic analysis of HWE kinases and GSR signaling described below.

5.3.3. E. litoralis DSM 8509 LovK is a photosensor

Like Caulobacter crescentus (282, 308), E. litoralis DSM 8509 possesses a single LOV-

HWE histidine kinase gene that is located adjacent to a single-domain response regulator gene on the chromosome (locus tags Ga0102493_111685 and 111686). We have named these genes lovK and lovR, respectively (Figure 6.33). The N-terminal LOV domain of LovK contains a complete flavin binding consensus sequence (276, 309) and a conserved cysteine residue required for light- dependent cysteinyl-flavin adduct formation (310). To validate that LovK indeed functions as a bona fide photoreceptor in vitro, we cloned and expressed LovK in a heterologous E. coli system, and purified the protein by affinity chromatography. LovK has a classic LOV domain visible absorption spectrum, with a lmax at 450 nm and vibronic bands at 425 nm and 475 nm. Illumination of the protein with blue light results in loss of these major bands, and a concomitant increase in absorption at 396 nm (Figure 6.34). This spectral signature is consistent with formation of a covalent cysteinyl-flavin C4(a) adduct upon illumination (310–313). We further expressed and purified LovK(C73A), in which the conserved cysteine postulated to form a flavin C4(a) adduct is mutated to a non-reactive alanine. LovK(C73A) retains flavin binding as evidenced by its absorption spectrum, but does not undergo the light-dependent bleaching of visible absorption bands that are indicative of cysteinyl-flavin adduct formation (Figure 6.34). These data provide evidence that E. litoralis DSM 8509 LovK is a photosensor that has the photochemical features of typical LOV proteins.

5.3.4. The E. litoralis DSM 8509 GSR regulon

Before testing whether LovK plays a role in the regulation of general stress response (GSR) transcription, we first sought to define the GSR regulon in E. litoralis DSM 8509 by RNA- sequencing (RNA-seq). To this end, we deleted the predicted orthologs of phyR (locus tag

Ga0102493_111538) and the nepR-ecfG operon (locus tags Ga0102493_111536 and 111535)

(Figure 6.33). As discussed in the introduction, these genes encode a protein partner switching system that controls activation of GSR transcription by sEcfG. Deletion of either phyR or ecfG were predicted to result in a strain incapable of activating GSR transcription. We compared the global transcriptional profiles of the DphyR and DnepR-ecfG deletion strains to wild type (WT) (Table

S3). To examine the effects of light, each strain was grown in cool white fluorescent light (60

µmol m-2 s-1) or in darkness (foil-covered tubes) for 24 hours prior to harvesting mRNA.

A shared set of 183 genes were differentially expressed (more than 1.5-fold; false discovery rate (FDR) p-value < 0.01) in DphyR and DnepR-ecfG compared to wild-type cells (Figure 6.35,

Table S4). With few exceptions, transcript levels in this gene set were lower in the two mutant strains compared to wild type indicating that these differentially regulated genes are transcriptionally activated by sEcfG in wild-type cells. We identified a predicted sEcfG binding motif that matched EcfG motifs from other Alphaproteobacteria (40–42, 314) in the promoter region of approximately one-quarter of the regulated genes (Figure 6.35, Table S4). Presence of these motifs predict the subset of genes in the E. litoralis GSR regulon that are directly regulated by the

ECF sigma factor, sEcfG.

Genes encoding transporters, cell envelope, and cell surface proteins are broadly represented in the E. litoralis DSM 8509 GSR regulon (Table S4). Transcription of genes encoding the ferritin-like protein DPS, a SOUL heme-binding protein, and superoxide dismutase

(sod) are commonly under the control of the GSR system in other Alphaproteobacteria (39, 222,

246, 286, 291, 315–321), and are regulated by nepR-ecfG and phyR in E. litoralis DSM8509

(Table S4). The promoters of dps, sodA, sodC, and a predicted SOUL heme-binding protein gene

(Ga0102493_112016) contain EcfG motifs and thus appear to be directly activated by sEcfG. Our transcriptomic data also provide evidence that expression of several DNA photolyases is activated by the GSR system.

Genes directly or indirectly regulated by sEcfG that may be relevant to the photobiology of

E. litoralis include a gene encoding a TspO/MBR tryptophan-rich protein (locus

Ga0102493_112559). This class of outer membrane protein has been reported to bind tetrapyrroles and promote their photooxidative degradation (322), and to negatively regulate photosynthesis in

Rhodobacter (323). Transcripts levels from this particular TspO/MBR gene are highly reduced in strains lacking phyR or nepR-ecfG. Unlike the fungus Aspergillus fumigatus, where TspO/MBR expression is strongly induced by light exposure (324), expression of E. litoralis TspO/MBR

(Ga0102493_112559) is lower in light than in dark conditions. Notably, steady-state transcript levels from the bacteriochlorophyll biosynthesis gene, bchC, are higher in both DphyR and in

DnepR-ecfG relative to wild type (Table S4). Transcripts of other bacteriochlorophyll biosynthesis genes including bchF, bchB and bchN are also higher in DphyR and DnepR-ecfG strains, though to a lesser extent than bchC (Figure 6.36). These results indicate that the E. litoralis GSR system directly or indirectly represses select genes involved in phototrophy.

5.3.5. Light-dependent regulation of transcription at the genome scale

In wild-type E. litoralis, transcripts in the GSR regulon are more abundant in dark-grown cells (aluminum foil-covered tubes) than in cells illuminated with white light (60 µmol m-2 s-1)

(Figure 6.35). In fact, measured GSR transcript levels from illuminated wild-type cells are only slightly higher than mutant strains lacking either phyR or nepR-ecfG. From this result, we conclude 95 that the GSR is only marginally active when cells are grown in continuous white light (Figure

6.35 and Table S4). A light versus dark difference in GSR transcripts is not observed in ∆phyR or

∆nepR-ecfG cells.

In wild-type cells, the majority of genes differentially expressed between light and dark conditions are in the GSR regulon. Of the 220 transcripts that change more than 1.4-fold (FDR p- value < 0.01), only 37 are not in the GSR regulon as defined (Figure 6.37, Table S5). Twenty- five of the differentially-expressed genes outside of the GSR regulon have similar light-dependent changes in the ∆phyR and ∆nepR-ecfG strains, and thus represent a GSR-independent light- regulated gene set. This GSR-independent gene set exhibits modest (less than 2-fold) regulation in response to white light treatment, and includes photosynthesis genes with reduced transcript levels in the light (Figure 6.36 and Figure 6.38). These genes are distinct from those derepressed in the GSR mutant strains (e.g. bchC) and include bchH, bchL, puhA, and bchM. We conclude that there are genes involved in photosynthesis (e.g. bch, puf, puh) for which expression is influenced by the GSR system, and others for which regulation is independent of GSR (Figure 6.36).

Reduction in photosynthesis gene expression in E. litoralis in the light is consistent with light repression of reaction center genes in the purple photosynthetic bacterium, Rhodobacter capsulatus (325), and downregulation of photosynthesis-related genes in the aerobic anoxygenic phototrophic bacterium, Dinoroseobacter shibae, upon a shift from heterotrophic growth in the dark to photoheterotrophic growth in the light (326). An operon of unknown function

(Ga0102493_112844-50) involved in acyl-CoA metabolism is repressed by light, independent of the GSR system. Eight metabolic genes of varying function are activated by light, independent of the GSR system (Figure 6.37).

5.3.6. Regulators of the GSR signaling pathway

Typically, transcription of genes encoding the core GSR regulators NepR, EcfG, PhyR and

GSR sensory kinases is activated by secfG (222, 246, 286, 291, 315, 316, 319, 320). In E. litoralis, a sEcfG-binding motif is positioned directly upstream of the nepR-ecfG operon. As expected, nepR- ecfG transcripts were reduced in the strain lacking phyR and in cells grown in the light (Figure

6.33C, Table S4).

It was previously noted that conserved nucleotides at -35 and -10 within the sEcfG-binding motif are nearly palindromic, which lead to the hypothesis that a palindromic site might function to drive bi-directional expression of oppositely oriented genes (314). The -35 and -10 sites upstream of nepR-ecfG have a strong palindromic character (GGAAC-N17-GTTCC) (Figure 6.38,

Table S4). Our transcriptomic data provide evidence that expression of both nepR-ecfG and the oppositely oriented HWE sensor kinase gene (Ga0102493_111537, gsrP) is activated from this single palindromic site. Both nepR-ecfG and gsrP are part of the GSR regulon as determined by

RNA-seq (Figure 6.33C, Table S4). Moreover, RNA-seq reads corresponding to nepR-ecfG and gsrP transcripts each begin 13-14 bp downstream of either end of this shared promoter motif

(Figure 6.38). While this is not a definitive mapping of transcriptional start sites, these data strongly suggest that this particular motif can functional bi-directionally. The mechanism by which sEcfG could regulate transcription from both strands at this site is not known. We note that the motif oriented toward nepR-ecfG has a stronger score in MEME, and that more RNA-seq reads mapped to nepR-ecfG than gsrP, which together suggest that sEcfG preferentially initiates transcription of nepR-ecfG. Finally, we observed a small number of RNA-seq reads that mapped to the negative strand of this motif and its downstream region (genome positions 619,592-619,621, Figure 6.38).

We speculate that these reads may correspond to transcription initiation from a cryptic or non-ECF

97 s promoter that would ensure some minimal level of nepR-ecfG expression under conditions in which the GSR was completely inactive.

In contrast to many previously described systems, an EcfG motif was not evident in the phyR promoter and, accordingly, transcript levels for this gene were not affected by deletion of nepR-ecfG or by light conditions (Figure 6.33C and Figure 6.38, Table S4). In other words, phyR appears to be constitutively expressed in E. litoralis DSM 8509. Constitutive expression of phyR likely necessitates the presence of a negative PhyR regulator to avoid constitutive activation of the

GSR.

HWE/HisKA2-family kinases are associated with the GSR regulatory system in

Alphaproteobacteria (40, 42, 327). E. litoralis DSM 8509 encodes five HWE/HisKA2-family kinases. Of these, only three are expressed at an appreciable level under our laboratory cultivation conditions, including: 1) the kinase encoded adjacent to nepR-ecfG operon (Ga0102493_111537),

2) lovK, and 3) an orphan sensor kinase (Ga0102493_11718) (Figure 6.33). Ga0102493_111537 encodes a transmembrane HWE sensor kinase with a periplasmic CHASE sensory domain; its transcription is activated by phyR and ecfG and by dark conditions (Figure 6.33C and Figure

6.38). As noted above, a shared EcfG motif identified between nepR-ecfG and this kinase likely promotes expression of both transcripts. The lovK-lovR promoter also contains an EcfG motif, and transcription of lovK-lovR requires nepR-ecfG and phyR (Figure 6.33C). The orphan HWE kinase,

Ga0102493_11718, lacks an EcfG motif in its promoter and is constitutively expressed in our experimental conditions (Figure 6.33C). Of the remaining HWE-kinase genes, transcripts corresponding to the orphan kinase gene Ga0102493_112963 are nearly undetectable (RPKM <

10). HWE-kinase gene, Ga0102493_111751, is encoded in an operon with a CRP-family

98 transcription factor and a single domain response regulator. Transcripts for these genes are also low abundance (Figure 6.39).

5.3.7. The role of three HWE-family sensor kinases in regulation of GSR transcription

To assess the role of HWE-family kinases 111537, 11718, and LovK in regulation of GSR transcription, we constructed strains bearing in-frame unmarked deletions of each kinase gene. We then evaluated GSR transcriptional output in these strains by measuring levels of a GSR-regulated transcript, dps, by qRT-PCR. This transcript was selected because 1) it exhibits a large dynamic range of expression between “GSR-ON” and “GSR-OFF” conditions (Figure 6.40), 2) it contains an EcfG motif in its promoter suggesting direct regulation by sEcfG (Table S4), and 3) primers to this transcript provided efficient amplification in one-step qRT-PCR reactions (see Materials and methods). Gene locus Ga0102493_112759, which encodes a methylmalonyl-CoA mutase, was selected as a control gene for normalization (Figure 6.40). In our RNA-seq data sets, transcripts for this gene were abundant (in the top 20% of all transcripts) and the RPKM coefficient of variation between samples was among the lowest, making this a suitable control gene for normalization. We tested these qRT-PCR primer sets in the same strains evaluated by RNA-seq and patterns of dps expression were consistent between these two methods (Figure 6.40).

In a strain lacking kinase 111537, steady-state levels of dps transcript increased in cultures grown both in the light and in the dark (Figure 6.41A). These data are consistent with 111537 functioning as a negative regulator of GSR transcription. Conversely, deletion of the 11718 resulted in decreased dps transcript levels, consistent with this gene functioning as a positive regulator of GSR transcription. Both of these transcriptional phenotypes were complemented by expression of the deleted gene from its native promoter on a replicating plasmid (Figure 6.42).

We have named these genes gsrP and gsrK, for general stress response phosphatase and kinase

99 respectively. We note that the putative phosphatase and/or kinase activity of these proteins has not been established, but our genetic data are consistent with these biochemical activities against

PhyR.

Deletion of the lovKR operon did not have a significant effect on dps transcript levels as measured by qRT-PCR (Figure 6.41A). Moreover, we did not observe large differences in GSR transcription in a ∆lovKR strain at the genome scale compared to wild type in our RNA-seq experiments (Figure 6.35). Nonetheless, GSR transcripts were uniformly lower in ∆lovKR compared to wild type when strains were grown in the dark (Figure 6.35D and E). The average log2 (fold change) difference in all GSR transcripts between these strains is -0.25, which reflects

~15% reduction in GSR transcription in the ∆lovKR strain (p < 0.0001). This trend is specific to the GSR regulon: the average log2(fold change) difference for all transcripts is 0.01 reflecting that, on average, the transcriptome at the whole-genome scale remains unchanged. These results provide evidence that lovKR plays a subtle role as an activator of GSR transcription in the dark.

To examine possible redundancy in the functions of these HWE kinases, we evaluated the effect of deleting pairs of kinases, leaving one of the three genes intact. In a ∆gsrK ∆lovKR double deletion strain where gsrP remains, dps transcript levels are lower than either of the single deletion strains in both the light and dark (Figure 6.41A). This result supports a model in which both GsrK and LovK function as GSR activators; moreover GsrP does not activate GSR transcription in the dark or in the light. In the ∆gsrP ∆lovKR double deletion strain where gsrK remains, dps transcripts are elevated in both the light and the dark, similar to the strain lacking only gsrP. Finally, in the

∆gsrP ∆gsrK strain where lovKR remains, dps transcripts are higher than wild type in both the light and the dark (Figure 6.41A) and are differentially expressed in response to light treatment.

In this strain, dps levels are higher in dark-grown than light-grown cells. To further investigate

100 whether LovK functions as a global activator of GSR, we measured the transcriptome of the ∆gsrP

∆gsrK strain grown in the light and in the dark by RNA-seq. These experiments confirm the single gene (dps) results in this strain. Specifically, GSR transcripts in this strain are broadly elevated compared to wild type in either light condition, and are higher in the dark than the light (Figure

6.35, Table S4). Together these data support a model in which LovK can function as an activator of GSR transcription that has enhanced activity in the dark.

To test whether LOV domain photochemistry (i.e. cysteinyl-flavin covalent adduct formation) contributes to the light-dark difference in LovK-dependent transcription, we mutated the conserved LOV domain cysteine (C73) to an alanine in the ∆gsrP ∆gsrK strain. Purified

LovK(C73A) mutant protein is blind to light (Figure 6.34), and in the ∆gsrP ∆gsrK lovK(C73A) strain, GSR transcription was no longer sensitive to light (Figure 6.41B). Specifically, dps levels were comparably high in light and dark grown cultures and similar to dark grown ∆gsrP ∆gsrK lovKR+ cultures. From these data, we again conclude that LovK is a more potent activator of GSR in its dark state.

We further tested whether the conserved histidine phosphorylation site in LovK (H161) was required to regulate GSR transcription. dps transcript levels in the ∆gsrP ∆gsrK lovK(H161A) strain are similar to a strain lacking all three kinases (∆gsrP ∆gsrK ∆lovKR) indicating that a lovK(H161A) mutant behaves like a ∆lovKR deletion. We conclude that LovK phosphorylation is necessary for LovK to activate GSR transcription (Figure 6.41B). Together, the data provide evidence that gsrP, gsrK and lovK coordinately regulate GSR transcription under our assayed conditions.

5.4. DISCUSSION

101

Regulation of transcription by alternative sigma factors including sS in

Gammaproteobacteria (11, 328), sB in select Gram-positive bacteria (35), and sEcfG in

Alphaproteobacteria (40, 42) confers general resistance to a range of physical and chemical stress conditions in vitro. Though these general stress response (GSR) s factors are conserved within phylogenetic groups, the input signals that cue their activation and the s-dependent transcriptional outputs vary across species. GSR inputs and outputs presumably reflect the distinct physicochemical challenges that particular species encounter within their niches. In this study, we report the development of the aerobic anoxygenic photoheterotroph (AAP), Erythrobacter litoralis

DSM 8509, as a comparative genetic model system to study regulation of GSR transcription by sEcfG. More specifically, we have sought to define the regulatory role of HWE-family sensor histidine kinases (280)— including a photoresponsive LOV-HWE kinase — in sEcfG-dependent transcription in E. litoralis.

5.4.1. Light, LOV, and bacterial stress responses

Proteins containing photosensory LOV domains are widely distributed in archaea, eukarya, and bacteria (275). The possibility that visible light regulates bacterial stress responses via LOV domains was first noted (273)after the discovery of the Bacillus subtilis LOV-STAS protein, YtvA, which functions as an activator of sB-dependent transcription (329). It was later shown that blue light can indeed activate sB-dependent transcription in B. subtilis via YtvA (330). Subsequent studies of a related LOV-STAS protein from Listeria monocytogenes support the conclusion that light regulation of sB by LOV proteins occurs more broadly in Gram-positive bacteria (331, 332).

A regulatory link between LOV and sS has also been reported in Pseudomonas syringae, where

102 white light was shown to repress expression of the gene encoding sS (rpoS) via a LOV histidine kinase (333).

Though the Alphaproteobacterial GSR sigma factor, sEcfG, is not related to sB or sS, studies of Caulobacter crescentus LovK provide evidence that LOV-HWE kinases repress sEcfG- dependent transcription. Experiments in Brucella abortus have identified a related LOV-HWE kinase, LovhK, that activates sEcfG-dependent transcription (286). However, these studies have not provided evidence that visible light is an input signal that influences sEcfG-dependent transcription

(246, 292). In E. litoralis DSM 8509, we have shown that the activity LovK as a regulator of sEcfG is influenced by light. More specifically, LovK functions with two additional HWE-family kinases,

GsrP and GsrK, to control GSR output (Figure 6.43A): GsrK is a strong GSR activator, GsrP is a repressor, and LovK is an activator that can be directly modulated by light (Figure 6.43A). These data contribute to an emerging model of GSR regulation in Alphaproteobacteria in which consortia of HWE/HisKA2-family kinases regulate sEcfG.

Our RNA-seq experiments show that E. litoralis GSR transcription is highly activated in the dark, and only marginally activated in cells continuously illuminated with white fluorescent light. In contrast, light transiently induces transcription of the core GSR regulators (Dshi_3834-

3837) and other stress response genes in the related AAP species, Dinoroseobacter shibae (326).

Similarly, blue light activates rpoE- and rpoH-dependent stress responses in the phototrophic

Alphaproteobacterium Rhodobacter sphaeroides (334, 335). We did not measure transient responses of E. litoralis to light or dark shifts, and the advantage (if any) of elevated GSR transcription in the dark is not known.

5.4.2. LovK functions as part of a consortium of GSR sensor histidine kinases

103

Although LovK can clearly function as a photoreceptor, it is not explicitly required for light/dark control of GSR transcription. Specifically, transcription from a GSR-regulated promoter remains light-responsive in a gsrK+ gsrP+ ∆lovKR strain. Thus, the combined activities of GsrK and GsrP must be balanced in a manner that leads to higher transcriptional output in dark relative to illuminated conditions (Figure 6.41A). This light responsiveness could emerge from enhanced activity of GsrK in the dark, enhanced repression by GsrP in the light, or a combination of both.

There is no evidence to suggest that photons are a direct signal for either sensor, thus we predict that either GsrK and/or GsrP indirectly sense a metabolic or physiological change that occurs upon shifts in the light environment.

We favor a model in which GsrP is a more potent GSR repressor in cells grown in the light

(Figure 6.43A). In our experimental conditions, gsrK alone strongly activates GSR transcription in light and dark (Figure 6.41A). This gene is constitutively expressed in both conditions (Figure

6.33C). While our results suggest that GsrK activity is not affected by light, we cannot entirely exclude the possibility that light conditions influence GsrK. GSR output from the strain bearing only gsrP is equivalently low in light and dark conditions (Figure 6.41A). While light may influence the activity of GsrP as a repressor, it is difficult to assess this possibility in the absence of a kinase that activates GSR transcription. sEcfG–dependent transcription of gsrP comprises a negative feedback loop that controls GSR transcriptional output; the level of gsrP transcripts is dramatically reduced in light conditions (Figure 6.33C, Table S4). This regulatory circuitry leads one to predict that the activity of GsrP as a repressor is inversely proportional to its steady-state levels in the cell. The activities of E. litoralis LovK, GsrK, and GsrP likely serve to both counteract and reinforce each other depending on environmental conditions. The possibility that E. litoralis and other Alphaproteobacteria can coordinately perceive multiple environmental inputs through a

104 consortium of HWE-family kinases may endow these species with the ability to execute complex decision-making processes with regard to control of GSR transcription.

Finally, we note that our data support a model in which LovK is active in both the light and dark, though activity of LovK as a positive regulator of sEcfG-dependent transcription is enhanced in the dark. These in vivo data are consistent with in vitro data showing that light does not regulate

E. litoralis HTCC 2594 LOV kinases in a binary (i.e. ON-OFF) manner, but rather modulates phosphorylation kinetics (290, 336). Though blue light typically enhances the enzymatic activity of LOV kinases in vitro, there is precedent for the unlit, dark state being the active state in vivo. In

D. shibae, a short LOV protein is a dark activator of pigment biosynthesis (337). In this case, the

LOV domain may be thought of as a dark sensor rather than a light sensor.

One may speculate on reasons why LOV proteins are often incorporated into GSR systems in the bacterial kingdom, but the fact remains that we have little understanding of the physiological relevance of LOV-regulated responses to changes in the light or redox environment in bacteria. E. litoralis DSM 8509 is a tractable experimental system with interesting physiological features. This species can now be leveraged as a comparative model to investigate core molecular mechanisms underlying environmental stress responses in Alphaproteobacteria.

5.5. MATERIALS AND METHODS

Growth of E. litoralis

Erythrobacter litoralis DSM 8509 was obtained from the American Type Culture Collection

(ATCC 700002). For these studies, this organism was grown in Difco Marine Broth 2216. When broth was prepared according to manufacturer’s instructions, cells tended to clump in flocs.

Dilution of marine broth to 0.5X reduced flocculation, and thus all cultures were grown at this

105 broth dilution (18.7 g of powder per L). 0.5X marine broth was sterilized by autoclaving and, prior to inoculation with E. litoralis, the medium was passed through a sterile 0.22 µm filter to remove precipitates. Liquid cultures were grown in an Infors air incubator at 30 ˚C in glass culture tubes, inclined at a 45˚ angle, shaking at 200 RPM. For “light” conditions, a bay of 10 fluorescent Philips and Sylvania T8 lights inside the shaking incubator was switched on. For “dark” conditions glass tubes were carefully wrapped in aluminum foil. For molecular genetic manipulation and isolation of mutants, colonies were grown on 0.5X marine broth solidified with 15 g of agar per L of medium. Agar plates were grown in a 30 ˚C air incubator, or at room temperature in ambient light.

Growth of Escherichia coli strains

E. coli strains were grown in LB Miller medium (10 g peptone, 5 g yeast extract, 10 g NaCl per L) at 37˚C. Growth medium was solidified with 1.5% agar. Antibiotics were added at the following concentrations as appropriate: kanamycin, 50 µg/ml; gentamycin 15 µg/ml; chloramphenicol 12

µg/ml.

Antibiotic sensitivity testing

To evaluate the appropriate antibiotic concentrations to use for selection in genetic manipulations,

E. litoralis DSM 8509 cells were first grown in a streak on 0.5X marine agar at 30 ˚C for 2-3 days.

Cells were scraped from the agar plate into 1 ml of 0.5X marine broth and evenly suspended by pipetting up and down. The optical density (OD) at 660 nm was was adjusted to ~ 0.1 AU. Cells were then 10-fold serially diluted and 20 µl of each dilution was spotted onto 0.5X marine broth agar that had been supplemented with a range of antibiotics concentrations. The antibiotic concentrations tested were as follows: 200, 100, 50, 25, 10 µg/ml for ampicillin, carbenicillin and tetracycline; 100, 50, 20, 10, 5, 1 µg/ml for gentamycin, spectinomycin, streptomycin, apramycin,

106 and rifampicin. After the liquid absorbed into the agar, the plates were incubated at 30 ˚C for 1 week. The minimal antibiotic concentrations that resulted in at least 4 orders of growth inhibition are reported in Table S2.

Molecular cloning for plasmid generation

Plasmids for expression or allele replacement were generated using routine techniques. Genes or loci of interest were PCR amplified with KOD Xtreme Hot Start polymerase (Millipore Sigma) using the primers listed in Table S6. To aid amplification of these GC-rich sequences, PCR reactions were supplemented with 5% DMSO (final concentration). Primers included a 5’ extension with restriction endonuclease sites or overhangs for overlap extension PCR reactions.

Null alleles were generated by first amplifying fragments ~500 bp upstream and ~500 bp downstream of the gene of interest. These two fragments were “stitched” together with overlap extension PCR to generate the desired allele. PCR products were cleaned using GeneJet DNA cleanup columns (Thermo Fisher), digested with appropriate restriction endonucleases (NEB) to generate overhang sequences, and ligated into similarly digested plasmids using T4 DNA ligase

(NEB). Ligation reactions were transformed into chemically competent TOP10 E. coli by heat shock. Transformants were selected on LB supplemented with the appropriate antibiotic and grown overnight at 37 ˚C. The inserted sequence and cloning junctions of all plasmids were confirmed by PCR amplification with plasmid specific primer sequences, treatment with ExoSapIT (Applied

Biosystems) followed by Sanger sequencing (University of Chicago Comprehensive Cancer DNA

Sequencing and Genotyping Facility). Plasmids generated for and used in this study are listed in

Table S6.

107

Transformation of E. litoralis

We evaluated and optimized two methods for introducing DNA into E. litoralis DSM8509, electroporation and conjugation. Replicating plasmids (with BBR or RK2 replication origins) could be introduced by electroporation, but this method was not efficient enough to introduce non- replicating (i.e. suicide) plasmids for chromosomal integration. Conjugation, by triparental mating, was used to introduce suicide plasmids containing oriT sequences, and also replicating plasmids containing oriT sequences. Gentamycin (10 µg ml-1) was used to select for clones carrying pBVMCS-4 derived plasmids. Chloramphenicol (1 µg ml-1) was used to select for clones carrying integrated plasmids, which permitted us to generate in-frame deletion or allele replacement strains.

Detailed electroporation and conjugation protocols follow.

Electroporation: Cells were scraped from a freshly grown plate that had been grown for 3 days at

30 ˚C. A pellet of 150-200 µl of cells was suspended in 750 µl 0.5X marine broth. Cells were pelleted by centrifugation (1 min at 14,000 x g) at 4 ˚C. The cell pellet was washed 3 times with

750 µl ice cold sterile water with a 1 min centrifugation step at each wash. After the final wash, the cells were resuspended with 140 µl cold water. The total volume of cells and water was ~200

µl. 60 µl cells were mixed with 0.1-1 µg purified plasmid and placed in a 1 mm electroporation cuvette. Cells were subjected to a single 1.8 kV pulse using the EC1 setting on a MicroPulser

(Bio-Rad). Time constants were ~ 4-5 msec. 450 µl of 0.5X marine broth was added to the cuvette and cells were transferred in this broth to a sterile 13 mm glass culture tube and incubated at 30 ˚C shaking at 200 rpm for 2-4 hours. After this outgrowth, cells were plated on selective medium

(about 200 µl per 100 mm petri dish). These plates were incubated at 30 ˚C for about 1 week.

After 5 days, pinprick-sized colonies were visible. After 7-8 days, colonies were picked and struck on fresh selective medium to confirm antibiotic resistance and to select away any non-transformed

108 cells. We confirmed that the clones carried the plasmid of interest by colony PCR, amplifying plasmid specific sequences using cells as the template.

Conjugation: Plasmids encoding oriT sequences were transferred from E. coli TOP10 donor strains to E. litoralis using a helper strain (MT607 / pRK600) (338) which encodes the pilus required to mobilize oriT containing sequences. E. litoralis recipient strains were inoculated from fresh plates into 2 ml 0.5x marine broth in 13 X 100 mm glass culture tubes and grown shaking at

30˚C overnight. The E. coli TOP10 donor strain carrying the plasmid to be transferred was grown in 2 ml LB supplemented with the appropriate antibiotic and the E. coli helper strain was grown in 2 ml LB supplemented with chloramphenicol at 37˚C overnight. The three cultures were mixed at a ratio of 1 ml recipient, 200 µl donor, 200 µl helper. The mixed cells were centrifuged for 1 min at 10,000-14,000 x g. The pellet was resuspended in 50-100 µl 0.5x marine broth and then spotted onto a fresh 0.5x marine broth agar plate. After the liquid in the spot of cells absorbed into the agar, the plate was incubated overnight at room temperature or 30 ˚C. The mix of cells was scraped from this non-selective plate and spread on 0.5X marine broth agar supplemented with 1) the appropriate antibiotic to select for acquisition of the plasmid and 2) 100 µg /ml nalidixic acid

(Nal) to counterselect against the E. coli donor and helper strains. E. coli strains exhibit increased resistance to Nal when grown on marine broth agar compared to LB, which required increasing the Nal concentration above typical concentrations used for counterselection. These selective plates were incubated for 7-8 days at 30 ˚C. Colonies that emerged were struck on fresh selective

0.5x marine broth agar plates and transformants were checked using the same approach described above for electroporated transformants.

109

Two-step chromosomal allele replacement

A standard two-step allele replacement approach with sucrose/sacB counterselection (339, 340) was optimized for E. litoralis DSM 8509. pNPTS138-derived allele replacement plasmids were introduced by conjugation as described above. Chloramphenicol resistant clones were inoculated into 2 ml 0.5X marine broth and allowed to grow without selection for 8-24 hours. These cells were spread on 0.5X marine broth agar supplemented with 7.5% sucrose to select for clones in which a second recombination event excised the plasmid. After approximately 1 week of growth at 30 ˚C, sucrose resistant colonies were replica patched on agar plates with or without chloramphenicol. From the chloramphenicol sensitive clones (i.e. clones in which the plasmid had excised) the locus of interest was PCR amplified and sequenced to distinguish clones bearing the parental allele from those bearing the desired new allele.

PacBio whole-genome sequencing

E. litoralis DSM 8509 genomic DNA was extracted using guanidium thiocyanate as previously described (341). Standard Pacific Biosciences (PacBio) large insert library preparation was performed. Briefly, DNA was fragmented to approximately 20kb using Covaris G tubes.

Fragmented DNA was enzymatically repaired and ligated to a PacBio adapter to form the

SMRTbell template. Templates larger than 10kb were BluePippin (Sage Science) size selected, annealed to sequencing primer, bound to polymerase (P6), bound to PacBio Mag-Beads and

SMRTcell sequenced using C4 chemistry. The genome was sequenced using two SMRT cells and was assembled de novo using HGAP3 and polished with quiver. This process yielded a single, closed contig. The genome was automatically annotated using the DOE-JGI pipeline (342).

HWE/HisKA_2-family histidine kinases and the core GSR regulators phyR, ecfG, and nepR were identified manually and annotated before sequence submission to GenBank. Raw PacBio reads

110 can be accessed through the NCBI sequence read archive at accession SRS1630618. The complete, annotated sequence of E. litoralis DSM 8509 is available at GenBank accession CP017057.

RNA extraction

For RNA-seq and qRT-PCR, RNA was extracted using a Trizol reagent based protocol. First, cells were struck from freezer stocks onto fresh 0.5X marine broth agar plates. After 3 days of growth, a scoop of mixed colonies of cells were inoculated into 2 ml 0.5X marine broth in 13 x 100 mm culture tubes and grown for 24 hours 30˚C shaking on a 45 degree incline at 200 RPM. Cultures were diluted to 0.005 OD660 in 2 ml fresh 0.5x marine broth and allowed to continue growth at 30

˚C shaking at 200 RPM. After 20-22 hours, these cultures were diluted to 0.001 OD660 in 6 tubes

(for RNA-seq) or 2 tubes (for qRT-PCR) with 2 ml each of 0.5X marine broth. Half of the tubes were placed in the top row of a shaker rack under a bay of 10 T8 fluorescent light bulbs (light conditions). The other half were wrapped in foil (dark conditions) and grown in the bottom row of the same shaker rack. After 22-24 hours of growth in constant light or constant dark conditions, culture densities reached 0.15-0.25 OD660. Cells were harvested by rapid centrifugation in small

“genotype-condition” batches to minimize handling time between culture growth and cell lysis.

For RNA-seq, the 3 x 2 ml from each genotype-condition set were distributed over 4 x 1.5 ml microfuge tubes and centrifuged for 60-90 seconds at 14,000-17,000 x g. After quickly aspirating the supernatant liquid, the four pellets were resuspended in a total of 1 ml Trizol reagent

(Invitrogen). For qRT-PCR, 1.5 ml of culture was harvested as above and resuspended in 1 ml

Trizol. Cells lysed in Trizol were immediately stored at -80 °C until extraction. For RNA-seq, five samples of each genotype-condition were collected, each grown on a different day.

To extract RNA from the Trizol suspended cells, samples were thawed and incubated at

65˚C for 10 minutes. After addition of 200 µl of chloroform, samples were vortexed, incubated

111 on the benchtop for 5 minutes, then centrifuged for 12-15 minutes at 14,000-17,000 x g. The aqueous phase (~500 µl) was transferred to a new 1.5 ml tube and 450 ul isopropanol was added to precipitate the nucleic acid. Samples were frozen overnight at -80˚C then centrifuged at 17,000 x g for 30 minutes at 4 ˚C. The pellet was washed twice with 750 µl cold 70% ethanol, suspended in RNase free water (200 µl for RNA-seq samples and 100 µl for qRT-PCR samples), and stored at -80 ˚C.

RNA-seq

RNA was subjected to DNase digestion using TURBO DNase I (Thermo Fisher). RNA samples were bound to an RNAeasy column (Qiagen) and digested on the column by application of 70 µL of DNase coctail (7 µL DNase, 7 µL 10x Buffer, 56 µL diH2O). Digests were incubated at room temperature for 45 minutes. RNA was cleaned and eluted from the column using buffers provided with the RNAeasy columns. DNA removal was confirmed by PCR using the RNA samples as a template and primers that amplify ~ 100 bp products (set 1: F-GACGGAGAAAAAGGCATCGC,

R-GATTCGCCGTGTTCATCTGC; set 2: F-CCCACGAACCGATTTCATGG, R-

CCTTCGGGGAGTTTCAAGCA). Absence of amplification confirmed removal of contaminating genomic DNA. Amplification from pre-DNAse samples served as a positive control. rRNA was depleted from the sample using the Gram-negative bacteria Ribo-Zero rRNA

Removal Kit (Illumina-Epicentre). RNA-seq libraries were prepared with an Illumina TruSeq stranded RNA kit according to manufacturer's instructions. The libraries were sequenced on an

Illumina HiSeq 4000 by the University of Chicago Functional Genomics Core Facility. Data were analyzed using the RNA-seq workflow in CLC genomics workbench v11.0. RNA-sequencing reads have been deposited in the NCBI GEO database under accession GSE126532.

112 qRT-PCR

The GSR-dependent transcript, dps (Ga0102493_111653), and an endogenous control transcript

(Ga0102493_112759) were evaluated using TaqMan probes and SuperScript III Platinum One-

Step qRT-PCR Kit (Invitrogen) with a QuantStudio 5 real time PCR system (Thermo Fisher). The primer-probe sets (dps F – TCTCATCGCCGAACTCAAC; dps R –

CGTGCCAGTGGAAATTCTTG; dps probe –

/5HEX/CTTCGCGCT/ZEN/GTTCACCAAGACC/3IABkFQ/) and (ctrl F –

AGATCGAAATGCTGTTGAAACG; ctrl R – GACCATCCAGAACGACATCC; ctrl probe -

/56-FAM/CCGCAACAC/ZEN/CTATATCTACCCGCC/3IABkFQ/) were custom prepared by

Integrated DNA Technologies. In control one-step RT-PCR reactions with a dilution series of template, both primer-probe sets exhibit 91-94% efficiency. Primers and probes were mixed to make a 10 µM F, 10 µM R and 5 µM probe stock solution. 0.4 µl primer-probe stock solution was used in each 20 µl reaction. 50 nM ROX served as the reference dye in each reaction. Each RNA sample was diluted to 2.5 ng µl-1; 4 µl were used in each 20 µl reaction. Each sample was assayed in triplicate for each probe set. The average Ct from these technical replicates was considered the

Ct for the sample. “No RT” control reactions were conducted on each sample with each probe to ensure that the signal from contaminating genomic DNA was less than 5 % of the signal (i.e. that the Ct from the “No-RT” reaction was at least 4 cycles later than the Ct from the “+RT” reaction.

Two control samples (WT – Dark, and WT – Light samples) were diluted to 2.5 ng µl-1 and frozen in 50 µl aliquots. One aliquot was thawed and assayed on each plate to ensure consistency between plates. Reaction parameters were: 50 ˚C 4 min, 95 ˚C 5 min, followed by 40 cycles of 95 ˚C 15 sec, 60 ˚C 30 sec where fluorescence was measured each cycle. For each sample, ∆Ct (Ct(dps)-

113

Ct(control)) was calculated. Then the average WT-dark ∆Ct was subtracted from each ∆Ct to generate a ∆∆Ct, which is the same as log2(fold change) compared to WT-dark.

Protein expression and spectroscopy

E. litoralis lovK (gene locus Ga0102493_111685) and a variant of lovK in which the flavin adduct- forming cysteine was mutated to an alanine (C73A) were cloned into the pET28a expression plasmid; see Table S6 for primers. Protein was expressed from these plasmids in E. coli

BL21(DE3). Briefly, 2 liters of LB (containing 50 µg ml-1 kanamycin) was inoculated with a 100 ml of an overnight culture. The culture was grown at 37°C / 220 rpm to OD600 ≈0.8 and induced with 0.5 mM IPTG for four hours before cells were pelleted by centrifugation. For purification, the cell pellet was resuspended in 50 ml of resuspension buffer (10 mM Tris (pH 7.6), 150 mM

NaCl, 10 mM imidizole) and lysed by two passages through a LV-1 microfluidizer. The lysate was then clarified by centrifugation (15 minutes at 35,000 x g) and applied to 2.5 ml of Ni-NTA resin on a gravity column. The resin had been pre-equilibrated with lysis buffer. The resin was washed with 5 column volumes (CV) of resuspension buffer, followed by a step gradient of buffer with increasing concentrations of imidazole. Specifically, the column was washed with 12 ml resuspension buffer containing 75 mM imidazole, 4 ml buffer with 200 mM imidazole, and finally

5 ml buffer with 500 mM imidazole for elution. The purity of the different fractions was assessed on a 12% SDS-PAGE gel. Peak fractions were desalted with a Zeba spin column (ThermoFisher) with a 7000 kDa MWCO. The visible absorption spectrum of purified LovK and LovK(C73A) was measured in a Tecan Spark 20M. The “lit” state of LovK was generated by illuminating purified protein with a panel of 96 LED bulbs (3 mm bulbs, 430 nm peak wavelength) held 10-20 cm from the sample for 30 seconds (until the visible yellow color was bleached).

114

5.6. SUMMARY

Extracytoplasmic function (ECF) sigma factors are a major class of environmentally-responsive transcriptional regulators. In Alphaproteobacteria the ECF sigma factor, sEcfG, activates general stress response (GSR) transcription and protects cells from multiple stressors. A phosphorylation- dependent protein partner switching mechanism, involving HWE/HisKA_2-family histidine kinases, underlies sEcfG activation. The identity of these sensor kinases and the signals that regulate them remain largely uncharacterized. We have developed the aerobic anoxygenic photoheterotrophic (AAP) bacterium, Erythrobacter litoralis DSM 8509, as a comparative genetic model to investigate GSR regulation. Using this system, we sought to define the contribution of visible light and a photosensory HWE kinase, LovK, to GSR transcription. We identified three

HWE kinase genes that collectively regulate GSR: gsrK and lovK are activators, while gsrP is a repressor. GSR transcription is higher in the dark than in light, and the opposing activities of gsrK and gsrP are sufficient to achieve light-dependent differential transcription. In the absence of gsrK and gsrP, lovK alone is sufficient to regulate GSR transcription in response to light. This regulation requires a photochemically active LOV domain in LovK. Our studies establish a role for visible light and HWE kinases in light-dependent regulation of GSR transcription in E. litoralis, an AAP species.

115

6. APPENDIX B – FIGURES

Figure 6.1 - Schematic representation of a growth curve The initial lag phase (highlighted in blue) depends of the conditions the cells were in when the measurements commenced (frozen, stationary, different media, etc.). In log phase (or exponential phase, green shading) cells are in the best condition to replicate, and are doubling as fast as the conditions permit. Once nutrients run out (or other factors intervene, like density, stress, accumulation of toxic metabolites, etc.) cell growth slows down and they roll over, reaching a plateau, where the net number of cells does not increase (stationary phase, ochre shading). Stationary phase will persist for an amount of time that depends on a multitude of factors, including the specific organism in question, but will eventually lead to the death phase (red highlight) where cells are actively dying. At the end of the death phase, a small amount of persister cells will endure in the long-term stationary phase (LTS, purple highlight).

116

Figure 6.2 - Growth of B. ovis harboring bcaA1-4BOV alleles in an unsupplemented atmosphere A) Cartoon illustrating the forward genetic selection for spontaneous mutants of Brucella ovis ATCC 25840 that grow without added CO2. Cells were grown on plates at 37 °C with 5% CO2 supplementation for 48 hrs then inoculated into BB and left in a shaking incubator at 37°C in standard atmospheric conditions (i.e. 0.04% CO2). Growth was monitored by cell culture density measurements (OD600) and individual colonies were isolated from tubes where growth was evident. B) Growth curve from three independent experiments of Brucella ovis ATCC 25840 (WT), ∆bcaA and B. ovis bcaA1BOV strains. Cells were grown either in 5% CO2 (top) or 0.04% CO2 ++ (bottom). C) Growth curve from three independent experiments with bcaABOV (bcaABOV ) or ++ bcaA1BOV (bcaA1BOV ) overexpressing strains. Cells were grown either in 5% CO2 (top) or 0.04% CO2 (bottom) after inducing expression from pSRK with 1mM IPTG. Strain carrying the empty vector (EV) was used as a control. D) Doubling time of Brucella ovis ATCC 25840 strains either carrying the pSRK empty vector plasmid (EV) or one of the four selected (“restored”) alleles ++ ++ ++ bcaA1BOV, bcaA2BOV, bcaA3BOV and bcaA4BOV (bcaA1BOV , bcaA2BOV , bcaA3BOV and ++ bcaA4BOV , respectively). Strains that did not grow are indicated (no growth). Experiment was

117

Continuation: Figure 6.2 – Growth of B. ovis harboring bcaA1-4BOV alleles in an unsupplemented atmosphere performed on separate days with technical replicates. Bar graphs indicate the standard deviation for 12 measurements per strain.

Figure 6.3 - A single nucleotide deletion at the 3’ end of bcaABOV enables B. ovis growth without CO2 supplementation A) Alignment of a segment of the 3’ end of bcaABOV (BOV_RS08635) from wild-type B. ovis and the 16 selected mutants that can grow without added CO2. Gaps show the single nucleotide deletions (highlighted teal dashes) in the bcaA locus. M1, M2, M3 and M4 are the four classes of single nucleotide deletions that we observed. Annotated nucleotide position is indicated at the top. B) Schematic representation of the wild-type B. ovis ATCC 25840 allele, bcaABOV (top), compared to the frameshifted mutant allele from the M1 cluster of mutants, bcaA1BOV (bottom). The site of the frameshift is indicated with a vertical arrow. Teal shading indicates the portion of the gene that has an altered coding sequence following the frameshift that results from deletion of a single guanosine at position 1,768,986. C) Multiple protein sequence alignment of the C-terminal

118

Continuation: Figure 6.3 – A single nucleotide deletion at the end of the 3’ end of the bcaABOV enables B. ovis growth without CO2 supplementation frameshift site of wild-type B. ovis ATCC 25840 BcaABOV and the four selected mutant alleles (M1 to M4). Teal shaded area highlights the frameshifted amino acids. Gray shaded area highlights sequence that is conserved in all five alleles. Boxed amino acids show protein sequence variation at the frameshift site (black arrows). D) Schematic of the bcaA locus in strains used for experiments in panel E; wild-type B. ovis ATCC 25480 bcaABOV allele (top), in-frame bcaABOV deletion allele (DbcaA) (middle), and a B. ovis strain in which the wild-type allele is replaced with the frameshifted bcaA1BOV allele at the native locus (bottom). PbcaA indicates expression from the native bcaA promoter. E) Bar graph of doubling times in minutes for the three strains in panel D grown with either 5% CO2 supplementation or in air without added CO2 (0.04%). Error bars represent standard deviation of replicates from three independent experiments (each performed with at least two technical replicates), for a total of 8 measurements per sample. Strains that failed to grow are indicated with “no growth”. **** indicates significance of p<0.0001, calculated using one-way ANOVA followed by Tukey’s post-test. F) Schematic of RK2-derived plasmids (pSRK) harboring two different bcaA alleles. pSRK carrying a lac inducible (Plac) wild-type bcaABOV (top) or frameshifted bcaA1BOV (bottom) were transformed into wild-type B. ovis ATCC 25840. ++ ++ Plasmid-bearing strains are referred to as bcaABOV and bcaA1BOV , respectively. G) Bar graph ++ of doubling times (in minutes) of strains carrying pSRK-bcaABOV (bcaABOV ), pSRK-bcaA1BOV ++ (bcaA1BOV ) or the empty vector (EV) as control. Cells were induced with 1mM IPTG, grown in BB with kanamycin to maintain plasmids, and cultivated in a standard air incubator (0.04% CO2) or with 5% CO2 supplementation. Error bars represent standard deviation of four independent experiments executed in triplicate (total number of measurements per sample = 12). Strains that failed to grow are indicated with “no growth”. **** indicates significance of p<0.0001, calculated using one-way ANOVA followed by Tukey’s post-test.

119

Figure 6.4 - Comparison of B. suis BcaA and E. coli b-carbonic anhydrases to BcaABOV and BcaA1BOV A) (Top) Multiple alignment of Brucella ovis BcaABOV against Brucella suis ATCC 23445 (BcaABSU) and Brucella suis 1330 (BcaABSU1330) homologs. The selected allele BcaA1BOV, which enables growth of B. ovis without added CO2, is included for comparison. Highlighted in black are the differences between each sequence and the genus-level consensus at top (see also Fig. 3 legend). (Bottom) Zoom-in on the C-terminal portion of the alignment that contains the bcaABOV frameshift present in wild-type B. ovis. A glutamine to arginine difference at position 204 of BcaABSU1330 is shaded in green. B) Multiple alignment of Escherichia coli MG1655 Can and CynT b-carbonic anhydrases with BcaABOV and BcaA1BOV. Orange asterisks show the conserved residues at the active site. Bottom right: table showing the amino acid identity between the proteins.

120

Figure 6.5 - Heterologous expression of two different Escherichia coli b-carbonic anhydrases enables growth of wild-type B. ovis ATCC 25840 without CO2 supplementation A) Schematic representation of RK2-derived plasmids (pSRK) transformed into either wild-type B. ovis ATCC 25840 (WT) or the in-frame bcaA deletion strain (DbcaA). pSRK plasmids carried ++ ++ either E. coli b-carbonic anhydrase can (can ) or cynT (cynT ) under inducible control of Plac. Empty vector (EV) plasmid was used as a control. B and C) Doubling time (in minutes) of strains outlined in A grown in 5% CO2 (B) or 0.04% CO2 (C) after induction with 1mM IPTG. Cells were grown in BB with kanamycin to maintain plasmids. The parent genotype (wild-type or DbcaA) is indicated at the top of each bar graph. The plasmid genotypes are indicated in each bar. Experiment was performed in triplicate two to four independent times (with two to three technical replicates for a total of 5 to 14 measurements per sample). Error bars represent standard deviation of at least two independent experiments. Strains that failed to grow are indicated with “no growth”.

121

Figure 6.6 - Sequence polymorphisms in BcaA orthologs across the genus Brucella A) Schematic of the BcaA protein sequence polymorphisms based on alignment of 774 BcaA orthologs from Brucella spp. available in the PATRIC database and found in Data Set 1, including B. ovis BcaA1BOV (teal; M1 cluster). Genus level consensus of all sequences is depicted on top. Species-specific consensus sequences were generated for species with at least 5 representatives (sequences not included in Fig. 5 are grouped in ‘Other, Data set 1). The species-specific consensuses were then aligned as shown. Amino acid differences between the species-specific and genus-level BcaA consensus sequences are indicated in black. Brucella abortus sequences were split into two groups, B. abortus* and B. abortus**, based on the presence or absence of a frameshift (and premature stop codon) resulting from a single cytosine insertion 339 nucleotides after the annotated start codon, respectively (see also Fig. 6 and Results). The number of strains compiled in each species-specific consensus sequence is shown in parentheses. The frameshifted BcaA1BOV allele that enables growth of B. ovis without added CO2 is included in the alignment (teal). B) Enlargement of green boxed area in (A). The C-terminal end of B. ovis BcaA1BOV (M1 cluster) is highlighted in teal; the site of the frameshift is indicated (green angled arrow).

122

Figure 6.7 - Analysis of sequence and function of B. abortus bcaA alleles in B. ovis ATCC 25840 A) Nucleotide alignment of B. abortus bcaA from 374 sequenced isolates starting at position +339 (codon 112). B. abortus bcaA sequences cluster into 8 groups based on nucleotide variation. Cluster number is indicated on the left and the number of sequences in each cluster is shown in parenthesis. Translated sequence for each codon is shown in light gray. Cytosine insertion that frameshifts the genus-level consensus (cluster 2) is highlighted gray. Gray box in cluster 1 (i.e. allele bcaABAB) highlights the premature stop codon arising from C insertion. Second-site, single- nucleotide deletions or double nucleotide insertions (highlighted in blue) that restore reading frame to consensus are highlighted for clusters 3 through 8. Cluster 7 represents the bcaA allele present in Brucella abortus strain 2308 (bcaABAB2308). Clusters 2 and 3 have the same amino acid sequence, but different nt sequence. B) Schematic of bcaA locus in B. ovis ATCC 25840 strains that were functionally assayed in C. From top: wild-type bcaABOV allele, the gain-of-function bcaA1BOV allele, the bcaABAB pseudogene (cluster 1), or the bcaABAB2308 allele (cluster 7), each expressed from the native promoter (PbcaA). C) Normalized B. ovis doubling times comparing WT (bcaABOV allele) to strains harboring bcaA1BOV, bcaABAB, or bcaABAB2308 in either high (5%) or low (0.04%) CO2. Error bars represent standard deviation of replicates from three to eight independent experiments for a total of 11 to 23 measurements per strain. **** indicates significance of p<0.0001, ns = non-significant, calculated using one-way ANOVA followed by Tukey’s post-test. Strains that failed to grow are indicated as ‘no growth’. D) Tabulation of gain-of-function bcaABAB mutants that arose in 4 independent selection experiments. Table shows experiment number (Exp#), number of times a specific allele (bcaA1BAB-3BAB) was isolated across the four independent experiments (#alleles), and cluster assignment for that allele (Cluster). Cluster designation as per panel A.

123

Figure 6.8 - RNA-seq experimental set up and measured gene expression changes in B. ovis bcaA1BOV upon CO2 downshift

124

Continuation: Figure 6.8 - RNA-seq experimental set up and measured gene expression changes in B. ovis bcaA1BOV upon CO2 downshift A) Schematic of sample treatment prior to RNA-seq. B. ovis wild-type and bcaA1BOV strains were inoculated in BB at OD600 of 0.05 and grown with 5% CO2 (6 tubes per strain, only one is shown for clarity). Once cells reached OD600 ~0.1, 3 tubes of B. ovis bcaA1BOV cells were moved to roller in an air incubator (0.04% CO2). Cells in all tubes were 5 then grown to OD600 ~0.2, and the three culture tubes of Brucella ovis ATCC 25840 (WT) in 5% CO2 and all six B. ovis bcaA1BOV (bcaA1BOV) cultures (3 from 5% CO2 and 3 from 0.04% CO2) were harvested at this point for RNA prep. Three remaining WT tubes were then moved to the air incubator (0.04% CO2) where they were incubated for 2.5 hours before harvesting for RNA preparation. B) Volcano plot for genes differentially expressed in the B. ovis bcaA1BOV strain upon downshift from 5% CO2 to 0.04% CO2. Analysis as described for wild-type B. ovis ATCC 25840 in Fig. 5B. No genes showed significant differential expression between these two CO2 conditions (FDR < 0.001 and log2 fold change > |1|). See also Data Set 2. C) R2 values calculated when plotting log10 Counts Per Million (CPM) of B. ovis ATCC 25840 (WT) against B. ovis bcaA1BOV in 0.04% or 5% CO2. Replicates (r1-3) are indicated. Low R2 values are highlighted in a yellow gradient, high R2 values in a teal gradient, legend at the bottom.

125

Figure 6.9 - Gene expression changes in wild-type B. ovis and B. ovis bcaA1BOV upon CO2 downshift from 5% to 0.04% A) Growth of wild-type Brucella ovis ATCC 25840 inoculated in BB and grown in 5% CO2. In log phase, cells were shifted to 0.04% CO2 (teal shading) for approximately 90 minutes then shifted back to 5% CO2. Lag phase is highlighted with gray striped shading. Experiment was performed three independent times and a representative experiment is shown. B) Volcano plot showing log2 fold change in gene expression versus log10 of the false discovery rate (FDR)- corrected p-value after downshift of wild-type cells from 5% to 0.04% CO2. Each dot represents a gene. Genes with FDR p-values < 0.001 and with log2 fold changes > |1| were considered statistically significant and are indicated as black dots. Due to floating point value constraints in calculation of p-values, the smallest possible FDR p-value we were able to calculate is 1 ´ 10-15. All lower FDR p-values were arbitrarily assigned a value of 1 ´ 10-17. Genes that were considered upregulated upon CO2 downshift are indicated on the right (teal shading) and genes that are downregulated upon downshift are indicated on the left (gray shading); statistically significant

126

Continuation: Figure 6.9 - Gene expression changes in wild-type B. ovis and B. ovis bcaA1BOV upon CO2 downshift from 5% to 0.04% regulated genes in each set are in parenthesis. C) Same growth experiment in (A) but with the B. ovis bcaA1BOV strain, which was selected to grow without added CO2. D) Absolute expression of all measurable transcripts, in log10 counts per million (CPM), from wild-type B. ovis ATCC 25840 in high (5%) CO2 compared to B. ovis bcaA1BOV in 5% CO2 (teal), B. ovis bcaA1BOV in low (0.04%) CO2 (black) or wild-type in low CO2 (gray). Each dot represents the mean CPM for a gene (see Data Set 1). R2 values were calculated (see Results and Fig. 7C) implementing linear regression using GraphPad Prism 8. CPM were calculated using CLC Genomics Workbench 11.0. E) Steady- state bcaA transcript levels by strain and condition as indicated. Error bars represent standard deviation calculated from two to three independent samples subjected to RNA-seq measurements.

Figure 6.10 - KEGG pathway assignment of genes regulated in wild-type B. ovis in response to CO2 downshift KEGG pathways are shown vertically on the left, with bars indicating the number of CO2-regulated genes assigned to each pathway. Genes significantly upregulated in response to CO2 downshift are on the right (teal bars), and genes significantly downregulated in response to CO2 downshift are on the left (gray bars). Significance thresholds follow those outlined in Fig. 8B legend. Pathways with less than 3 genes, or where the number of upregulated versus downregulated genes was not

127

Continuation: Figure 6.10 - KEGG pathway assignment of genes regulated in wild-type B. ovis in response to CO2 downshift significantly different (see Materials and methods), were removed from graph. Select KEGG categories associated with the stringent response are marked by black arrows. Teal arrow marks the “Biotin metabolism” pathway.

Figure 6.11 - Heat map representation of a subset of genes regulated by CO2 downshift in wild-type B. ovis B. ovis gene accession numbers and RefSeq annotations are shown to the right of the heatmap. Genes were grouped into three categories: biotin biosynthesis, biotin-dependent enzymes and urease genes. Each column represents an individual RNA sample from either B. ovis ATCC 25840 (WT) or B. ovis bcaA1BOV grown in 5% or downshifted to 0.04% CO2, measured by RNA-seq. Strain replicates are indicated as r1, r2, and r3. The four genes that encompass the acetyl-CoA carboxylase complex and pyruvate carboxylase are highlighted in teal. ure pseudogenes are marked with an asterisk; the urea transporter, which is a pseudogene in B. ovis, is in bold. Z-scores express the number of standard deviations from the mean for each gene (see Materials and methods).

128

Figure 6.12 - Nucleotide differences between sets of functional genes and pseudogenes in B. ovis ATCC 25840 and their B. abortus ATCC 2308 orthologs normalized by gene length Each dot represents a gene. Genes are cataloged in Table 5. Bars represent the mean and standard error of the mean for each group. Percentage of synonymous mutations in urease and T4SS genes are indicated.

129

Supplemental figure 1

OD600 0.05 0.12 0.9 2.4 BOV_RS01000 ABC transporter ATP-binding protein BOV_RS06660 tRNA (adenosine(37)-N6)-dimethylallyltransferase MiaA BOV_RS00995 thiaminase II BOV_RS04130 phosphoribosylformylglycinamidine synthase subunit PurL BOV_RS04155 phosphoribosylaminoimidazolesuccinocarboxamide synthase BOV_RS03500 phosphoribosylglycinamide formyltransferase BOV_RS00690 bifunctional uridylyltransferase/uridylyl-removing protein BOV_RS02130 phosphoribosylamine--glycine ligase BOV_RS02280 amidophosphoribosyltransferase BOV_RS03505 phosphoribosylformylglycinamidine cyclo-ligase BOV_RS08585 bifunctional phosphoribosylaminoimidazolecarboxamide formyltransferase/inosine monophosphate cyclohydrolase BOV_RS07900 hypothetical protein BOV_RS08630 lytic murein transglycosylase BOV_RS09045 tRNA (guanosine(37)-N1)-methyltransferase TrmD BOV_RS06065 alpha/beta hydrolase BOV_RS08265 5-(carboxyamino)imidazole ribonucleotide mutase BOV_RS09735 tRNA uridine-5-carboxymethylaminomethyl(34) synthesis enzyme MnmG BOV_RS09740 tRNA uridine-5-carboxymethylaminomethyl(34) synthesis GTPase MnmE BOV_RS10110 phosphate ABC transporterpermease protein PstA BOV_RS04795 RluA family pseudouridine synthase BOV_RS01785 saccharopine dehydrogenase BOV_RS06060 serine O-acetyltransferase BOV_RS04630 ATP-dependent helicase BOV_RS05445 L D-transpeptidase BOV_RS03110 AMP nucleosidase BOV_RS09535 heavy metal translocating P-type ATPase BOV_RS09615 D-alanyl-D-alanine carboxypeptidase BOV_RS15320 elongation factor 4 BOV_RS04930 DUF922 domain-containing protein BOV_RS03230 GTP pyrophosphokinase rsh BOV_RS05340 HAMP domain-containing histidine kinase BOV_RS06975 alpha/beta hydrolase BOV_RS14345 type I restriction endonuclease subunit R BOV_RS02060 tRNA1(Val) (adenine(37)-N6)-methyltransferase BOV_RS04850 glutathione-disulfide reductase BOV_RS01700 large conductance mechanosensitive channel protein MscL BOV_RS00245 L D-transpeptidase BOV_RS06915 N-acetylmuramoyl-L-alanine amidase BOV_RS04900 MCE family protein BOV_RS10170 PhoH family protein BOV_RS00145 putative sulfate exporter family transporter BOV_RS00475 dihydroxy-acid dehydratase BOV_RS01830 (S)-ureidoglycine aminohydrolase BOV_RS05655 MerR family DNA-binding transcriptional regulator BOV_RS06655 acetolactate synthase 3 large subunit BOV_RS05540 RIP metalloprotease RseP BOV_RS06685 protease modulator HflC BOV_RS10675 hypothetical protein BOV_RS00460 heme exporter protein C BOV_RS13135 cytochrome c biogenesis protein CcdA BOV_RS02505 hypothetical protein BOV_RS07830 disulfide bond formation protein B BOV_RS05580 DUF1849 domain-containing protein BOV_RS03700 RNA degradosome polyphosphate kinase BOV_RS03540 L-lactate permease BOV_RS06675 PDZ domain-containing protein BOV_RS09555 cobaltochelatase subunit CobS BOV_RS10595 amino acid permease BOV_RS07420 DUF58 domain-containing protein BOV_RS02300 class I SAM-dependent methyltransferase BOV_RS07425 DUF4159 domain-containing protein BOV_RS07430 hypothetical protein BOV_RS15415 cardiolipin synthetase BOV_RS08905 hypothetical protein BOV_RS08900 heme biosynthesis protein HemY BOV_RS14115 acyl-CoA dehydrogenase BOV_RS09550 cobaltochelatase subunit CobT BOV_RS13960 glutamine synthetase BOV_RS09365 TIGR02302 family protein BOV_RS09775 maf-like protein BOV_RS11600 ribulokinase BOV_RS12590 thiol reductant ABC exporter subunit CydD BOV_RS10720 oligoendopeptidase F 3.4 BOV_RS04905 hypothetical protein BOV_RS15345 NAD(P)-dependent oxidoreductase BOV_RS07415 MoxR family ATPase 3.0 BOV_RS08595 NAD-glutamate dehydrogenase BOV_RS14750 histidine utilization repressor 2.6 BOV_RS08845 phosphoenolpyruvate--protein phosphotransferase BOV_RS08345 hypothetical protein BOV_RS07285 peptidoglycan editing factor PgeF 2.2 BOV_RS10290 peptidase P60 BOV_RS00895 methionine synthase 1.8 BOV_RS04615 glycosyltransferase family 2 protein BOV_RS01715 glycosyl transferase BOV_RS02700 mannose-1-phosphate guanylyltransferase/mannose-6-phosphate isomerase 1.4 BOV_RS07170 glycosyltransferase family 1 protein phosphomannomutase BOV_RS07180 DUF1176 domain-containing protein 1.0 BOV_RS10190 XRE family transcriptional regulator BOV_RS01370 DUF2948 domain-containing protein Increasing Fitness Score 0.6 BOV_RS05330 Trk system potassium transporter TrkA BOV_RS03070 hypothetical protein BOV_RS10970 ChbG/HpnK family deacetylase 0.2 BOV_RS10975 glycosyltransferase family 39 protein BOV_RS14735 D-amino acid dehydrogenase small subunit -0.2 BOV_RS01585 two-component sensor histidine kinase BOV_RS01125 sugar ABC transporter substrate-binding protein BOV_RS02550 MBL fold metallo-hydrolase -0.6 BOV_RS13775 glycine dehydrogenase BOV_RS13785 glycine cleavage system aminomethyltransferase T -1.0 BOV_RS06590 PhzF family phenazine biosynthesis protein BOV_RS10125 phosphate regulon transcriptional regulatory protein PhoB BOV_RS07865 RluA family pseudouridine synthase -1.4 BOV_RS12140 FadR family transcriptional regulator BOV_RS14100 helix-turn-helix transcriptional regulator -1.8 BOV_RS03705 exopolyphosphatase BOV_RS10070 SMP-30/gluconolactonase/LRE family protein BOV_RS09630 NAD(P)-dependent oxidoreductase -2.2 BOV_RS11585 sugar ABC transporter ATP-binding protein BOV_RS11580 ABC transporter substrate-binding protein -2.6 BOV_RS11590 sugar ABC transporter permease BOV_RS11815 class 1b ribonucleoside-diphosphate reductase subunit alpha Decreasing Fitness Score BOV_RS01680 prolyl aminopeptidase -3.0 BOV_RS10060 orotidine-5'-phosphate decarboxylase BOV_RS12310 glycerol kinase -3.4 BOV_RS13210 aspartate carbamoyltransferase BOV_RS13215 dihydroorotase Figure 6.13 - Fitness profile of B. ovis transposon insertion mutants as a function of growth phase Heat map of fitness scores of 118 genes with t-like significance score ³ |4|. Genes were grouped by hierarchical clustering. Each column is a point during the growth curve (OD600, indicated at the

130

Continuation: Figure 6.13 - Fitness profile of B. ovis transposon insertion mutants as a function of growth phase top of the heatmap). Each row is a gene. Gene locus tags and annotated functions are indicated on the left.

Figure 6.14 - Assessment of B. ovis mutant strain fitness as a function of growth phase identifies cysE as a determinant of stationary phase fitness A) Heat map showing mean fitness scores (n = 3) of B. ovis mutant strains harboring barcoded transposon insertions in non-essential genes. Each row represents a gene, and each column is a point during the growth curve in BB (OD600). Genes with a t-like significance score ³|4| and fitness value ³|1| in at least one timepoint are included in the heat map. Genes were hierarchically clustered (left), which yielded four main groups (right). The fitness profile of strains harboring RB 131

Continuation: Figure 6.14 - Assessment of B. ovis mutant strain fitness as a function of growth phase identifies cysE as a determinant of stationary phase fitness Tn-himar insertions in cysE is marked with an asterisk on the right of the heat map. B) Alternative representation of data shown in A, where the average fitness score for each mutant (gene) is plotted as a function of timepoint. Individual lines represent genes, shaded area delimits the max and minimum values within that group. cysE is presented as a thick blue line. Group 1, blue; Group 2, red; Group 3, yellow; Group 4, green.

Figure S2 A 3% 3% 3% 5% 21% 5% 5% 5% 5%

10% 8% 45%

10% 19% 11%

20% 11% Group1 11% Group2

RNA Metabolism, Modification, and Translation (6) Purine Metabolism (10) Amino Acid Metabolism (4) 25% Peptidoglycan, Envelope, Membrane (4) 33% General (10) DUF and Hypothetical Proteins (7) 75% 67% Cobalamine, Heme, and Cytochromes (4) Signalling and Transcriptional Regulation (7) Group3 Group4 Sugar Metabolism (1) Pyrimidine Metabolism (1) Glutathione Metabolism (1) Transport and Channels (8) DNA modification and Cell Cycle (1) B

Average fitness score Annotated product Locus ID Group 0.050.12 0.9 2.4

serine O-acetyltransferase BOV_06060 - 0.18 - 0.11 - 1.65 - 3.31 1

AMP-nucleosidase BOV_03110 - 0.79 - 1.16 - 0.99 - 3.06 1

bifunctional (p)ppGpp synthetase/guanosine-3',5'-bis(diphosphate) 3'-pyrophosphohydrolase BOV_03230 - 1.52 - 0.18 - 1.78 - 2.65 2

saccharopine dehydrogenase family protein BOV_01785 - 0.2 - 0.32 - 1.22 - 2.45 1

MerR family DNA-binding transcriptional regulator BOV_05655 - 1.33 - 1.29 - 1.72 - 2.43 2 alpha-beta hydrolase BOV_06975 - 0.52 - 0.93 - 1.73 - 2.21 2

Figure 6.15 - Functional classification of B. ovis genes whose disruption significantly impacts fitness during growth in Brucella broth A) Gene function pie chart of the four clusters of genes identified in Fig. 2, grouped based on predicted function. Area of each pie chart is proportional to the number of genes in that group.

132

Continuation: Figure 6.15 - Functional classification of B. ovis genes whose disruption significantly impacts fitness during growth in Brucella broth DUF = Domain of Unknown Function. B) Subset of genes from Fig. 2A with fitness score values ³ |2| in stationary phase.

Figure 6.16 - ∆cysE enters stationary phase prematurely; this growth defect is rescued by addition of cysteine to the growth medium A) Schematic of cysteine biosynthesis from serine. Teal arrows indicate cysteine biosynthesis enzymes annotated in the Brucella ovis genome (RefSeq accessions NC_009505 and NC_009504). Enzymes: CysE (BOV_RS06060, cysE, O-acetylserine transferase); CysK1 (BOV_RS09280, cysK , cysteine synthase A). B) Representative growth curves of wild type (WT, circles), and ∆cysE (triangles) with (gray and ochre, respectively) or without (black and teal, respectively) addition of 4 mM cysteine (Cys) to the growth medium. C) Representative growth curves of WT carrying the pSRK empty vector (EV, black circle) or pSRK-cysE (gray square), and ∆cysE carrying pSRK (teal triangle) or pSRK-cysE (purple diamond) in BB with 1 mM IPTG, 50 µg/ml Kan. Error bars represent standard deviation of technical replicates in representative experiments.

133

Figure 6.17 - The stationary phase phenotype of a ∆cysK1 ∆cysK2 double deletion phenocopies ∆cysE and is rescued by cysteine A) Growth of wild type (black circles), ∆cysK1 (black diamonds), ∆cysK2 (black squares) and ∆cysK1 ∆cysK2 (teal triangles) in BB. B) Growth of the same four strains in A with 4 mM cysteine added to the broth. C) Growth curves of WT carrying the pSRK empty vector (black circles), and ∆cysK1 ∆cysK2 carrying either pSRK-cysK1 or pSRK-cysK2 (purple triangles) grown with 50 µg/ml Kan and 1 mM IPTG. Error bars represent standard deviation of technical replicates and may be smaller than symbol size. Growth curves were conducted at least three independent times. A representative curve is shown for each.

134

Figure 6.18 - B. ovis ∆cysE is sensitive to H2O2 treatment; ∆cysE growth defect and peroxide sensitivity is mitigated by glutathione A) Schematic of glutathione metabolism. Enzymes annotated in the Brucella ovis genome (RefSeq accessions NC_009505 and NC_009504) are indicated in bold: GshA (BOV_RS13935, glutamate—cysteine ligase), GshB (BOV_RS10075, glutathione synthase), and Gor (BOV_RS04850, glutathione disulfide–reductase). GSH (glutathione, reduced state); GSSH (glutathione disulfide, oxidized state). B) Growth of wild type (circles) or ∆cysE (triangles) in BB with (gray and ochre, respectively) or without (black and teal, respectively) 4 mM GSH added to the medium. Error bars represent standard deviations and may be smaller than symbol size. C) Hydrogen peroxide survival assay showing the log2 ratio of CFU of treated (20 mM H2O2) versus untreated (mock PBS) cultures. Black bars represent wild type and teal bars represent ∆cysE strains. Addition of either 4 mM cysteine (+ Cys) or 4 mM GSH (+ GSH) to the medium is indicated by the shaded boxes. GSH and cysteine was washed away from the culture prior to peroxide treatment. D) Same assay as in C but with strains carrying plasmids to test genetic complementation. Strains were treated with 15 mM H2O2. Black bars represent WT carrying an empty vector (WT/ pSRK-EV), teal bars represent ∆cysE carrying an empty vector (∆cysE / pSRK- EV), and purple bars represent complemented ∆cysE (∆cysE / pSRK-cysE). p-values: * p < 0.05; ** p < 0.01; *** p < 0.001, **** p < 0.0001, calculated using one-way ANOVA (followed by Dunnett’s multiple comparison test, to ∆cysE in C or ∆cysE/ev in D). Hydrogen peroxide survival assay as in C, comparing wild type B. ovis (black bars) to the ∆cysK1 ∆cysK2 double deletion strain (teal bars) F) Same as in D but with strains carrying plasmids to test genetic complementation. Black bars represent B. ovis carrying the empty vector pSRK-EV (WT/EV), teal

135

Continuation: Figure 6.18 - B. ovis ∆cysE is sensitive to H2O2 treatment; ∆cysE growth defect and peroxide sensitivity is mitigated by glutathione bars represent ∆cysK1 ∆cysK1 with pSRK-EV (∆∆/ev); and purple bars represent the ∆cysK1 ∆cysK2 strain carrying either pSRK-cysK1 (∆∆/1, dark purple) or pSRK-cysK2 (∆∆/2, light purple). Error bars represent standard error of the mean for 3 or 4 independent experiments.

Figure 6.19 - B. ovis ∆cysE has reduced fitness in the intracellular niche of human macrophage-like cells and an ovine testis epithelial cell line Log10 colony forming units (CFU) per well of WT (black circles) or ∆cysE (teal triangles) isolated from infected THP-1 (A and C) or OA3.ts mammalian cells (D) enumerated at 2, 24 and 48 hours (hrs) post infection (p.i.) (A and D) or at 2, 4, 8, 12, and 24 hrs p.i. (C). B) Log10 CFU/well per well of WT (black circles) or ∆cysK1 ∆cysK2 (teal triangles) strains isolated from infected THP-1 enumerated at 2, 24 and 48 hrs p.i. p-value comparing recovered B. ovis 24 hrs p.i. were calculated using an unpaired t-test A, B and D. * p < 0.05; *** p < 0.001, **** p < 0.0001. Infections were repeated 3 -5 independent times. Error bars represent standard error within the representative experiment.

136

Figure 6.20 - Recovered CFUs decrease at 72 hrs post infection Colony forming units of wild type B. ovis (WT, black circles) and ∆cysE (teal triangles) cells recovered from infected THP-1 cells at 2, 24, 48, and 72 hours post infection (Hrs p.i.). Error bars represent SEM for three independent experiments. Figure S4 AB THP-1 OA3.ts 7 6 WT/ev WT/ev ∆cysE/ev ∆cysE/ev ∆cysE/cysE 6 ∆cysE/cysE **** 5 ** 5 *** *

4 Log(CFU/well)

3 3 0244802448 Hrs p.i. Hrs p.i.

Figure 6.21 - Genetic complementation of THP-1 and OA3.ts infection Log10 colony forming units (CFU) per well of WT/pSRK-EV (WT/ev, black circles), ∆cysE/pSRK-EV (∆cysE/ev teal triangles), and ∆cysE/pSRK-cysE (∆cysE/cysE, purple diamonds) isolated from infected THP-1 (A) or OA3.ts mammalian cells (B). Significance at the 24 hr time point was assessed using one-way ANOVA (Tukey’s multiple comparison test). Asterisks indicate p-values compared to ∆cysE or ∆cysE/ev: * p < 0.05; ** p < 0.01; *** p < 0.001, **** p < 0.0001. Infections were repeated 3-5 independent times. Error bars represent standard error within the representative experiment.

137

Figure 6.22 - B. ovis ∆cysE is more attenuated in THP-1 human macrophage-like cells than in ovine testis OA3.ts cells Ratio of recovered CFUs of B. ovis ∆cysE relative to WT at 24 hrs post infection (p.i.) in macrophage-like human THP-1 cells and in the sheep testis epithelial cell line, OA3.ts. p-value was calculated using a two-tailed unpaired t-test, ** p<0.01. Error bars represent SD of four independent experiments in each cell line.

Figure 6.23 - B. ovis ∆cysE strain is attenuated in Raw 264.7 macrophages Colony forming units of wild type (black circles) and ∆cysE (teal triangles) B. ovis cells recovered after infection of Raw 264.7 murine macrophages. p-value calculated using an unpaired t-test for the 24 hr time point. Error bars represent SEM for four independent repetitions of the experiment.

138

Figure 6.24 - Brucella ovis and B. ovis bcaA1 have no difference in morphology Contrast microscopy images showing wild type B. ovis (left) and B. ovis bcaA1 (right) picked from a 48 h-old plate grown in 5% CO2 supplementation. See Materials and methods from (244) (B. ovis eipA depletion assays) for additional information.

Figure 6.25 - Modelling of BcaA structure in B. ovis and comparisons to experimental structures of other carbonic anhydrases A) Left: modelling of BcaA1BOV based on the structure of b-carbonic anhydrase from Pisum sativum. In teal are the regions affected by the inactivating frameshift, in green the three amino acids that coordinate to the Zinc2+ ion at the active site, detailed on the right. Models of BcaA1 monomer were built as follows. Amino acid sequences were implemented in XtalPred (http://ffas.burnham.org/XtalPred-cgi/xtal.pl). Search of PDB homologs yielded 53 sequences.

139

Continuation: Figure 6.25 - Modelling of BcaA structure in B. ovis and comparisons to experimental structures of other carbonic anhydrases The most similar sequence (35% identical) for BcaA1 was from Pisum Sativum (PDB: 1ekj). The PDB file of the pea b-carbonic anhydrase sequence was fitted to adapt the BcaA1 sequence using Phyre2 {Kelley, 2015 #210} and Swiss-Model {Waterhouse, 2018 #211}. The resulting PDB file was imported in PyMOL (the PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC) for graphic rendering. B) Table showing the similarities between the carbonic anhydrases in Pisum sativum (Pisum), Synechocystis sp. (strain PCC 6803 / Kazusa) (Synch.), and Escherichia coli (E. coli) to B. ovis BcaA1 and E. coli CynT. The structures for the P. sativum, Synechocystis (PDB: 5swc), and E. coli Can (PDB:1t75) carbonic anhydrases are shown in C, D, and E, respectively. The metal ions and coordinated amino acids are shown in green, the ribbons corresponding to the regions that would be affected by the frameshift inactivating BcaA1BOV are shown in teal.

Figure 6.26 - BcaABOV does not purify in a soluble form and is mainly found in inclusion bodies E. coli Rosetta strains carrying pET28a-Hix(6x)-bcaABOV, pET28a-His(6x)-bcaA1, and pET28a- His(6x)-bcaABSU were inoculated in LB and grown at 37 °C, until OD600 » 0.6 was reached. Expression of N-terminally His-tagged proteins was induced with 1 mM IPTG for 4-6 hours. Cells were centrifuged and pellet was collected, resuspended in 50 ml of Binding buffer (10 mM Tris- HCl pH 7.4, 150 mM NaCl, 10 mM Imidazole) supplemented with DNAse A (1 µl/ml), PMSF (4 µl/ml) and half a tablet of complete protease inhibitor mix (Roche Applied Science), and lysed using a microfluidizer (Microfluidics LV1). The lysate was clarified by centrifugation. The soluble fraction (Sol.) and insoluble fraction (inclusion bodies, I.B.) were run on a 14% SDS-page gel (A). Purification of the His-tagged proteins was achieved through nickel affinity chromatography (nitrilotriacetic acid resin). The lysate (L) was run through the column, and the flow through (FT) was collected as well. The two wash steps (W1 and W2) were done using Wash Buffer (10 mM

140

Continuation: Figure 6.26 – BcaABOV does not purify in a soluble form and is mainly found in inclusion bodies Tris-HCl pH 7.4, 150 mM NaCl, 75 mM imidazole) and the bound proteins eluted twice (E1 and E2) in elution buffer 1 (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 200 mM imidazole) and a third time (E3) in elution buffer 2 (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 500 mM imidazole). Samples from the various purification steps were collected and run on a 14% SDS-page gel (B).

Figure 6.27 - BcaB from B. abortus cannot rescue B. ovis growth in 0.04% CO2 A) Amino acid alignment comparison between BcaB from Brucella abortus ATCC 2308 (BcaBBAB), Brucella ovis ATCC 25840 (BcaBBOV), and Brucella suis 1330 (BcaBBSU1330). Amino acids that do not match the consensus (based on Blosum 90 score matrix) are indicated in varying shades of gray as explained by the legend in the bottom right corner. B) Strains built for experiment in C: WT B. ovis (WT), B. ovis bcaA1BOV (bcaA1BOV), B. ovis or B. ovis bcaA1 carrying the B. abortus bcaB allele replacing bcaBBOV at the native locus (bcaBBAB and bcaA1BOV bcaBBAB, respectively). C) Doubling time normalized to WT B. ovis of strains grown in atmospheric Continuation: Figure 6.27 – conditions (0.04% CO2, right) or supplied with 5% CO2 (left). p-value was calculated using one way Anova and Tukey’s multiple comparison test, ns: non-significant, *

141

Continuation: Figure 6.27 – BcaB from B. abortus cannot rescue B. ovis growth in 0.04% CO2 p-value < 0.05, ** p- value < 0.01, **** p-value < 0.0001. See 2.5 Materials and methods for additional information on bacterial strains and growth conditions, plasmid and strain construction, growth curves, and sequence alignments.

Figure 6.28 - Deletion of cysE leads to differences in coloring and tolerance to different metals A) Wild type Brucella ovis (WT), B. ovis ∆cysE clone #1 (#1), or B. ovis ∆cysE clone #2 (#2) cells were scraped off of a 48 hour-old (top) or 96 hour-old (bottom) plain TSA (TSA) or TSA + 5% sheep blood plate (Blood). B) B. ovis carrying pSRK-EV (EV) or B. ovis ∆cysE carrying pSRK- EV (∆/ev) or pSRK-cysE (∆/cysE) cells scraped off of TSA + 5% sheep blood plates supplied with 50 µg/ml of kanamycin and 1 mM IPTG (bottom) or no IPTG (top). C) Disc diffusion assay plates were sterile cotton discs were imbued with different concentrations of either CdCl2 or CuSO4 (bottom cartoon) and placed on lawns of B. ovis (WT) or B. ovis ∆cysE (∆cysE). Pictures were taken after 3 days of spreading and 9 days. Asterisks indicate single colonies of CuSO4 resistant cells and white arrows highlight metallic shade around the CdCl2 10 mM disc on lawns of B. ovis ∆cysE cells. See 3.5 Materials and methods for details on strain building and additional information.

142

Figure 6.29 - Supplementation of methionine does not restore ∆cysE growth to wild type levels Growth of WT B. ovis (WT, circles) or B. ovis ∆cysE mutant strain (∆cysE, triangles) with cysteine (orange), methionine (green), or without supplemental additions to the media (black). See 3.5 Materials and methods for details on strain building, growth curves, and additional information.

143

Figure 6.30 - Predicted cysteine metabolism in B. ovis Illustration of the predicted molecules and enzymes involved in cysteine metabolism in B. ovis. Abbreviations: achy, adenosylhomocysteinase; chaC, glutathione-specific g- glutamylcyclotransferase; CoA, coenzyme A; cysC, adenylylsulfate kinase; cysE, serine-O- acetyltransferase; cysH, 3’-phosphoadenlylysulfate reductase; cysJI, sulfate reductase; cysK1/2, cysteine synthase; cysND, sulfate adenylyltransferase; gshA, glutamate—cysteine ligase; gshB, glutathione synthetase; metA, homoserine transsuccinylase; metC, cystathionine b-lyase; metH, methionine synthase; metK, S-adenosylmethionine synthase; metZ, O-succinylhomoserine

144

Continuation: Figure 6.30 – Predicted cysteine metabolism in ovis sulfhydrylase; mlgV, O-acetylhomoserine (thiol)-lyase; pepA, aminopeptidase A; pepN, aminopeptidase N.

Figure 6.31 - Similarities of conserved regions of OCBS and OASS to B. ovis CysK1/2, and effect of cystathionine supplementation Conserved regions identified by Devi and colleagues (257) in cysteine synthase (i.e. O-acetylserine sulfhydrylase, OASS), cystathionine b-synthase (CBS), and O-acetylserine-dependent CBS (OCBS) compared to CysK1 and CysK2 from B. ovis. Significantly different amino acids are underlined.

HWE/HisKA2 Histidine Kinases

NepR σ Rec σ P

PhyR PhyR~P

NepR EcfG RNAP σEcfG σ GSR transcription

EcfG EcfG inhibited activated

Figure 6.32 - Core regulators of the Alphaproteobacterial general stress response In Alphaproteobacteria, the general stress response (GSR) is often controlled by an ensemble of HWE/HisKA_2 family sensor histidine kinases that modulate the phosphorylation state of PhyR. PhyR phosphorylation promotes a protein partner switch that sequesters the anti-sigma factor, NepR, thereby releasing the extracytoplasmic function (ECF) sigma factor, EcfG, to initiate

145

Continuation: Figure 6.32 – Core regulatoris of the Alphaproteobacterial general stress response transcription of genes in the general stress response regulon. The core protein partner switch genes, phyR, nepR and ecfG are broadly conserved in Alphaproteobacteria. The number of kinases controlling the pathway is variable.

gsrP lovK lovR gsrK ecfG nepR phyR Ga0102493_111535-38 Ga0102493_111685-86 Ga0102493_11718 500 bp

B GAF Chase LOV PAS IM σ REC REC σ HWE HWE HWE HK HK HK

C ecfG nepR gsrP phyR lovK lovR gsrK 2500 2500 120 500 500 120 600

2000 2000 100 400 400 100 500 80 80 400 1500 1500 300 300 60 60 300

RPKM 1000 1000 200 200 40 40 200 500 500 20 100 100 20 100 N/A N/A N/A 0 0 0 0 0 0 0 L L L L L L L L L L L L L L L L L L L L L D D D D D D D D D D D D D D D D D D D D D WT ∆nepR ∆phyR WT ∆nepR ∆phyR WT ∆nepR ∆phyR WT ∆nepR ∆phyR WT ∆nepR ∆phyR WT ∆nepR ∆phyR WT ∆nepR ∆phyR ∆ecfG ∆ecfG ∆ecfG ∆ecfG ∆ecfG ∆ecfG ∆ecfG

Figure 6.33 - GSR regulators and their expression in E. litoralis DSM 8509 A) Genes encoding the GSR regulators in E. litoralis. The core partner switch genes (phyR, nepR, and ecfG) together with a HWE kinase gene (gsrP) constitute the GSR locus (left). The lovKR operon and gsrK are not adjacent to each other or to the GSR locus on the chromosome. Gene locus numbers (GenBank accession CP017057) are listed below each gene diagram. Predicted EcfG-binding motifs are marked with green circles at the base of the transcription start arrows. B) Domain structure of the proteins encoded by the genes in (A). Colors match the gene colors. GsrP is predicted to encode a transmembrane histidine kinase and is depicted in the inner membrane (IM). All other proteins lack transmembrane domains and are predicted to be cytoplasmic. C) RNA-seq analyses of transcript abundance from the genes in (A). Reads per kilobase per million reads (RPKM) of each gene is plotted for wild-type (WT), ∆nepR-ecfG or ∆phyR strains grown in the dark (D - hashed bars) or the light (L – light grey bars). Mean ± s.d. is plotted. n=5 per treatment; individual values from experimental replicates are dots.

146

A 0.10 LovK dark lit

0.05 Absorbance units

0.00 320 360 400 440 480 520 560 Wavelength (nm)

B LovK(C73A) 0.15 dark lit

0.10

0.05

0.00 320 360 400 440 480 520 560 Wavelength (nm) C Light minus dark 0.04 LovK 0.02 LovK(C73A)

0.00

-0.02

-0.04

∆ absorbance units-0.06 Absorbance units

-0.08 320 360 400 440 480 520 560 Wavelength (nm)

Figure 6.34 - Absorption spectroscopy provides evidence that LovK protein can function as a photosensor A-B) Visible absorption spectrum of purified E. litoralis DSM 8509 (A) LovK and (B) LovK(C73A) proteins. Lit spectrum was collected immediately after illuminating the protein with blue light. The dark spectrum is the same protein that was not illuminated. C) Light minus dark difference absorption spectrum for LovK (blue line) and LovK(C73A) (yellow line). Photoexcitation of the wild-type protein results in a signature loss of absorbance between at 440 and 480 nm and an increase in absorbance at 396 (310, 313) due to cysteinyl-flavin C4(a) adduct formation (311, 312). The LovK(C73A) mutant protein lacks the conserved cysteine required for adduct formation and does not exhibit light induced changes in absorbance.

147

A DARK LIGHT B ∆nepR-ecgG ∆phyR

gsrP gsrP

∆ ∆ 59 183 20 gsrK gsrK phyR lovKR phyR lovKR nepR-ecfG nepR-ecfG ∆ ∆ ∆ ∆ ∆ ∆ WT ∆ WT ∆ C 2

1 bits G G G T G C C C CCC C C C AAG C C C C C T G A C C A G C GT TC G TT CAC C C G C A G G G C CGT T A G A T CAC A G GA G G G T G TT A G G C C G G A G A T C A T A AAT T T G A T T T A G G G T T T TT A TC A AT C TT GT CGG A G CA CCA 0 G CAA G CGA AA GA A G 1 2 3 4 5 6 7 8 9 A 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 T26 T27 G28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 G-35A -10 D 2

∆gsrK∆gsrP-D 0 WT-D ∆lovKR-D

-2 ∆gsrK∆gsrP-L ( fold change ) 2 -4 log

compared to WT dark ∆lovKR-L Dark Light WT-L WT -6 ∆lovKR ∆gsrK∆gsrP ∆nepR-ecfG-L ∆nepR-ecfG -8 ∆nepR-ecfG-D

E DARK LIGHT **** **** 2 ns 1 0

0 -2 ( fold change )

2 -4 -1 log compared to WT dark log2(fold change) -6 gsrP scale gsrK gsrK gsrP 3.0 lovKR ∆ lovKR ∆ 0.0 0.6 1.2 1.8 2.4 WT ∆ ∆ WT ∆ ∆ -3.0 -2.4 -1.8 -1.2 -0.6

Figure 6.35 - General stress response regulon of E. litoralis DSM 8509 A) Heatmap showing relative expression of genes in the GSR regulon, defined by RNA- sequencing. Each row represents a gene, and each column represents a genotype-condition pair. Wild-type (WT) and mutant strains grown in the dark are on the left and strains grown under constant white light illumination (~60 µmol m-2 s-1) are on the right. The color or each block represents the log2(fold change) for an individual gene, where fold change is the ratio of RPKM(indicated condition) / RPKM(WT dark). Color scale is below the heat map. The heat map contains a 148

Continuation: Figure 6.35 – General stress response regulation of E. litoralis DSM 8509 subset of 148 genes in Table S4 with a fold change > 2 in the ∆nepR-ecfG or the ∆phyR data sets, and a max group mean RPKM > 10. GSR regulatory genes that were deleted in these strains were excluded from the gene set presented in the heatmap. B) Venn diagram of the differentially regulated genes in the ∆nepR-ecfG or ∆phyR strains compared to wild type. The criteria for genes deemed to be differentially regulated was a fold change > 1.5 with a false-discovery rate (FDR) p- value < 0.01. These genes in are listed in Table S4. C) Logo of the ECF s-type motif identified in the promoter region of a subset of genes in the GSR regulon; the logo was generated using MEME motif discovery tools. The -35 and -10 boxes (underlined) of this E. litoralis motif are consistent with previously described EcfG motifs (40, 42, 314). D) Relative expression of the 148 genes in panel A. Each position on the x-axis represents a gene ranked (left to right) by log2(fold change) in ∆nepR-ecfG dark relative to WT dark. Color-coded dots at each position represent the relative expression of a gene in different strain-condition combinations compared to WT dark (see key). Relative expression in dark-grown cultures are closed circles; light-grown cultures are open circles. E) Relative expression values for genes in the GSR regulon in WT, DlovKR and DgsrKDgsrP strains (grown in light or in dark) presented in a single column. Expression of each gene in these strain backgrounds is plotted relative to WT dark. The mean difference between each column was statistically assessed by one-way ANOVA followed by Dunnett’s multiple comparison test; **** indicates p<0.0001. The key in D corresponds to both graphs in E.

149

Light / Dark

-D / WT-D gsrP

-D / WT-D ∆ gsrK phyR lovKR nepR-ecfG nepR-ecfG phyR ∆ ∆ ∆ WT ∆ ∆ ∆ Ga0102493_112559, tryptophan-rich sensory protein Ga0102493_111604, protochlorophyllide reductase Ga0102493_111605, TonB-dependent receptor Ga0102493_111606, 3-phytase Ga0102493_111607, divinyl chlorophyllide a 8-vinyl-reductase Ga0102493_112207, hypothetical protein Ga0102493_111842, carotenoid 1,2-hydratase Ga0102493_111841, 1-hydroxycarotenoid 3,4-desaturase, CrtI Ga0102493_111840, demethylspheroidene O-methyltransferase Ga0102493_111839, 3-hydroxyethyl bacteriochlorophyllide a dehydrogenase, BchC Ga0102493_111838, chlorophyllide a reductase subunit X, BchX Ga0102493_111837, chlorophyllide a reductase subunit Y, BchY Ga0102493_111836, chlorophyllide a reductase subunit Z, BchZ Ga0102493_111835, light-harvesting complex 1 beta chain, PufB Ga0102493_111834, light-harvesting complex 1 alpha chain, PufA Ga0102493_111833, photosynthetic reaction center L subunit, PufL Ga0102493_111832, photosynthetic reaction center M subunit, PufM Ga0102493_111829, geranylgeranyl reductase, BchP Ga0102493_111828, MFS transporter, BCD family, chlorophyll transporter Ga0102493_111827, chlorophyll synthase, BchG Ga0102493_111826, transcriptional regulator PpsR Ga0102493_111825, Methanogenic corrinoid protein MtbC1 Ga0102493_111824, 3-vinyl bacteriochlorophyllide hydratase, BchF Ga0102493_111823, light-independent protochlorophyllide reductase subunit N, BchN Ga0102493_111822, light-independent protochlorophyllide reductase subunit B, BchB Ga0102493_111821, magnesium chelatase subunit H, BchH Ga0102493_111820, light-independent protochlorophyllide reductase subunit L, BchL Ga0102493_111819, magnesium-protoporphyrin O-methyltransferase, BchM Ga0102493_111818, MFS transporter, BCD family, chlorophyll transporter Ga0102493_111817, photosynthetic reaction center H subunit, PuhA Ga0102493_111816, PH domain-containing protein Ga0102493_111814, hypothetical protein Ga0102493_111813, magnesium-protoporphyrin IX monomethyl ester (oxidative) cyclase, AcsF

Light / Dark GSR log2(fold change) scale 0.0 0.5 1.0 1.5 2.0 -2.0 -1.5 -1.0 -0.5

Figure 6.36 - Relative expression of genes involved in phototropy Heatmap represents the same comparisons and same color scaling as in Figure 6.37. Genes were selected based on annotated functions in bacteriochlorophyll synthesis, phototrophy, or proximity to such genes. Genes are ordered by position on the chromosome. Genes in red text meet the cutoff criteria for inclusion in the GSR regulon (Table S4). Genes in blue text meet the cutoff criteria for inclusion in the light-dark regulon (Table S5).

150

AB Light / Dark D / WT-D

gsrP

Light / Dark D / WT-D

∆

D / WT-D phyR nepR-ecfG phyR lovKR gsrK nepR-ecfG ∆ ∆ ∆ ∆ ∆ WT ∆ gsrP

D / WT-D ∆ nepR-ecfG phyR lovKR gsrK nepR-ecfG phyR WT ∆ ∆ ∆ ∆ ∆ ∆ Locus Ga0102493_11xxxx, annotated function 1817, photosynthetic reaction center H subunit, puhA 1818, MFS transporter, BCD family, chlorophyll transporter 1820, light-independent protochlorophyllide reductase subunit L, bchL 1821, magnesium chelatase subunit H, bchH 1829, geranylgeranyl reductase, bchP 1835, light-harvesting complex 1 beta chain, pufB 1836, chlorophyllide a reductase subunit Z, bchZ 1837, chlorophyllide a reductase subunit Y, bchY 1838, chlorophyllide a reductase subunit X, bchX 2207, hypothetical protein 2844, 2,4-dienoyl-CoA reductase 2845, anthraniloyl-CoA monooxygenase 2846, enamine deaminase RidA 2847, acyl-CoA dehydrogenase 2848, enoyl-CoA hydratase/carnitine racemase 2849, DNA-binding transcriptional regulator, MarR family 2850, NADP-dependent 3-hydroxy acid dehydrogenase YdfG 2734, adenosylhomocysteinase 1274, S-adenosylmethionine synthetase GSR dependent light-dark regulation 1609, methylenetetrahydrofolate reductase (NADPH) 1608, ArsR family transcriptional regulator 2766, ATP-dependent RNA helicase RhlE 1677, cold shock protein (beta-ribbon, CspA family) 1678, hypothetical protein 455, membrane fusion protein, cobalt-zinc-cadmium e!ux system

log2(fold change) scale GSR independent 0.0 0.5 1.0 1.5 2.0 -2.0 -1.5 -1.0 -0.5

Figure 6.37 - The E. litoralis light-dark regulon overlaps with the GSR regulon A) Heatmap of differentially expressed genes in wild-type (WT) E. litoralis DSM 8509 grown in constant ~60 µmol m-2 s-1 white light or in the dark (D). Cutoff criteria for differential expression are fold change > 1.4 and FDR p-value<0.01 (see Table S5 for list of genes). Heatmap includes only genes with max mean RPKM >10. For each gene (in rows), the log2 (RPKMlight / RPKMdark) is shown for each genotype (columns). In addition, relative change in the ∆nepR-ecfG and ∆phyR strains compared to wild type (all grown in the dark) is presented to highlight congruence between light-regulated genes and the GSR regulon. B) Heatmap highlighting genes at the bottom of the cluster in panel A. These genes exhibit light-dependent regulation, but are not in the GSR regulon. Color scale corresponds to both panels.

151

A ecfG nepR gsrP phyR 16384 4096 1024 256 64 16 WT dark 4 WT light 1 618,500 619,000 619,500 620,000 620,500 621,000 621,500 622,000

B 300 EcfG binding site

200

100 Read depth Read depth

0 CTGGCACCTGGGAGGCGCGTGCTGTGGGAACGTAGCGCGAGGTGTTTGGTTCCGTCCACACACCAAGGCAAATGCTTTCCCAATGAACTCT

619,580 619,590 619,600 619,610 619,620 619,630 619,640 619,650 619,660 619,670 Genome position

Figure 6.38 - Palindromic sEcfG binding site lies between nepR-ecfG and gsrP A) RNA-seq reads mapped to the GSR locus from wild-type cells grown in the dark or in the light. Number of mapped reads are plotted as a function of genome position (GenBank accession CP017057). The genes in this region are colored as in Figure 6.33. The single sEcfG motif in this region is marked by a green dot and beige bar. Mapped read depth for phyR is similar in light and dark conditions, while RNA-seq reads are more abundant in the dark for nepR-ecfG and gsrP. B) Expansion of the intragenic region between nepR-ecfG and gsrP. Palindromic bases of the -35 and -10 sites of EcfG motif are dark green, and the intervening bases are lighter green. The beginning of the peak of reads for each transcript is marked with a vertical dashed line. The predicted +1 sites are colored and marked with an arrow to indicate the direction of transcription (colors correspond to the genes as in panel A).

152

GAF-(PAS)2-HWE CRP-TX (PAS)3-HWE SDRR 500 bp Ga0102493_112963 Ga0102493_111750-52

GAF PAS CRP-family PAS PAS sensory PAS HTH- PAS REC DNA BD HWE HWE HK HK

Ga0102493_11 Ga0102493_11 Ga0102493_11 Ga0102493_11 120 120 120 120 2963 1750 1751 1752 100 100 100 100 80 80 80 80 60 60 60 60 RPKM RPKM RPKM RPKM 40 40 40 40 20 20 20 20 0 0 0 0 L L L L L L L L L L L L D D D D D D D D D D D D WT ∆nepR ∆phyR WT ∆nepR ∆phyR WT ∆nepR ∆phyR WT ∆nepR ∆phyR ∆ecfG ∆ecfG ∆ecfG ∆ecfG

Figure 6.39 - Gene structure, protein domain structure and RNA-seq expression values for two additional unnamed HWE kinases encoded in the DSM 8509 genome Features are drawn as in Figure 6.33.

153

ABGSR-regulated target gene control gene Ga0102493_111653 Ga0102493_112759 dps, DNA starvaon/staonary phase protecon protein methylmalonyl-CoA mutase 4096 1 4096 3 log 2048 0 log 2048 2

2 2 (fold change) (fold change) 1024 -1 1024 1 512 -2 512 0 RPKM RPKM 256 -3 256 -1 128 -4 128 -2 64 DL DLDL DL DL 64 D LDLDL D L D L WT ∆nepR ∆phyR ∆lovKR ∆gsrP WT ∆nepR ∆phyR ∆lovKR ∆gsrP -ecfG ∆gsrK -ecfG ∆gsrK lovKR lovKR C qRT-PCR (Ct target - Ct control) scaled to WT dark 1

-1

(fold change) (fold -2 2

log -3 D L D L D L D L D L D L control WT ∆nepR ∆phyR ∆lovKR ∆gsrP -ecfG ∆gsrK lovKR

Figure 6.40 - Target and control genes used for qRT-PCR analysis of GSR transcription A) RPKM values for dps (a GSR-regulated gene), extracted from RNA-seq experiments (Table S3) plotted on a log2 scale. Bars represent mean ± s.d. of 5 independent samples (dots). Log2 (fold change) relative to wild type (WT) dark is scaled on the right y-axis. B) RPKM values for the methylmalonyl-CoA mutase gene used as the endogenous control gene for normalization plotted as in (A). C) qRT-PCR analysis of the same genotype-condition combinations assayed by RNA- seq in (A) and (B). These data are extracted from the same set of experiments presented in Figure 6.38 and are presented here for direct comparison to panel (A). As in Figure 6.38, each measurement represents the (Ctdps – Ctcontrol)sample – average (Ctdps – Ctcontrol)WT-D. This results in a log2(fold change) compared to WT-dark. Strains were grown and assayed together. Each sample was assayed in triplicate. Points represent the average value for each sample. Bars represent mean ± s.d. of the samples in each condition. As an internal control, the same wild type-dark and wild type-light samples were assayed on every plate (control).

154

A Dark 1 Light

-1 (fold change) 2

log -2 dps -3

gsrP + + ∆ + + + ∆ ∆ gsrK + + + ∆ + ∆ + ∆ lovKR + + + + ∆ ∆ ∆ + control wild single kinase mutants double kinase mutants type

B Dark 1 Light

-1 (fold change) 2

log -2 dps

-3

gsrP + + ∆ ∆ ∆ ∆ ∆nepR ∆phyR gsrK + + ∆ ∆ ∆ ∆ -ecfG lovKR + + + C73A H161A ∆ control wild lovK mutant alleles core GSR type pathway mutants

Figure 6.41 - Combinatorial control of GSR transcription by three HWE-family sensor histidine kinases A-B) qRT-PCR quantification of the levels of, dps, a GSR-regulated transcript. dps transcript measurements were carried out on RNA isolated from (A) a set of single and double HWE kinase mutant strains and (B) lovK mutant, ∆nepR-ecfG and ∆phyR strains grown in the dark (D, dark grey bars) or the light (L, light grey bars). Each measurement represents the (Ctdps – Ctcontrol)sample – average (Ctdps – Ctcontrol)WT-D, which yields a log2(fold change) in dps level relative to WT dark. Strains in each panel were grown and assayed together. Each sample was assayed in triplicate. Points represent the average value for each sample. Bars represent mean ± s.d. of all samples in each condition. As an internal control, a pair of wild type-dark and wild type-light samples was assayed on every plate (control).

155

-1

-2 (fold change) 2 -3 log

-4 condion DL DL DL DL DL DL DL DL DL DL DL DL plasmid none EV EV + EV + EV + EV + EV + genotype wild type ∆nepR-ecfG ∆phyR ∆gsrP ∆gsrK ∆lovKR

Figure 6.42 - Complementation of GSR transcription defects in strains with single deletions of GSR regulators qRT-PCR analysis of dps expression as a measurement of GSR transcription. All strains, including wild type (WT), carry a replicating plasmid and were grown in the presence of gentamycin to select for plasmid maintenance. Plasmids either carried the deleted gene under the control of its endogenous promoter (+) or were empty vectors (EV) to control for plasmid and selection effects. Grey shaded bars highlight mutant strains carrying the empty vector. The chromosomal genotype of the strains in each colored block is indicated at the bottom. All strains were grown in the dark (D – striped bars) or in the light (L – open bars). Data are presented as in Figure 6.38 and Figure 6.40C. The internal control RNA samples assayed on every plate in this experiment were from wild type samples that did not carry a plasmid (far left bars).

156

A Kinase contributions DARK LIGHT hυ

GsrK GsrP LovK GsrK GsrP LovK

GSR GSR transcription transcription ON OFF

B Signal wiring negative feedback

positive constitutive P expression feedback

NepR σEcfG

σEcfG RNAP GSR GSR transcription output

Figure 6.43 - Model of GSR regulation in E. litoralis DSM 8509 A) Three HWE-family sensor histidine kinases (LovK, GsrK, and GsrP) control output of sEcfG- dependent transcription. The weight of arrows shows the relative regulatory contribution of each sensor kinase under light and dark conditions. B) Model of GSR control network including protein interactions (solid lines) and transcriptional output (dashed lines). Most, but not all, of the GSR pathway regulators are subject to feedback regulation. PhyR, which serves to integrate signals, and GsrK, the primary activating kinase, are not under transcriptional control of sEcfG, and are instead constitutively expressed. This feature of the GSR system in E. litoralis DSM 8509 likely necessitates a negative regulator to prevent constitutive activation of GSR transcription; GsrP fulfills this function.

157

7. APPENDIX C – TABLES

Table 7.1 - Growth of independent spontaneous mutants derived from forward genetic a selection grown in an air incubator (0.04% CO2) after isolation OD at OD at WGS Exp.b Samples 600 600 24hrs 48hrs Samplesc WT1 0 0.001 Bov Parent WT2 0 0 Plate 1 0.011 1.645 Mutant 01 liq1 2 0.026 1.558 Mutant 02 3 0.01 1.654 Mutant 03 4 0.015 1.664 5 0.014 1.654 6 0.022 1.655 Mutant 14 7 0.007 1.64 liq2 8 0.014 1.657 Mutant 05 9 0.005 1.617 Mutant 04 10 0.009 1.646 11 0.01 1.647 12 0.012 1.655 Mutant 06 13 0.044 1.633 Mutant 07 14 0.02 1.639 Mutant 08 15 0.018 1.627 16 0.027 1.628 liq3 17 0.012 1.617 Mutant 09 18 0.015 1.609 19 0.18 1.642 Mutant 15 20 0.012 1.629 21 0.26 1.606 Mutant 10 22 0.021 1.651 Mutant 11 23 0.024 1.593 24 0.026 1.554 Mutant 12 25 0.038 1.635 Mutant 13 liq4 26 0.027 1.635 27 0.028 1.635 Mutant 16 28 0.068 1.638 29 0.028 1.634 a -5 Cultures were inoculated at an OD600 of 1.5 ´ 10 bSpontaneous independent mutants from independent experiments (Exp.). “Plate” refers to selection conducted on solid media instead of liquid, by transferring plated cells from 5% to 0.04% CO2 incubators. Liquid (liq) 1-4 are four independent experiments were cells were inoculated in broth. See Results and Materials and methods. “Bov” is the parent strain (Brucella ovis ATCC 25840) where two independent tubes were inoculated for comparison with the mutants. cSamples indicated were sent for whole genome sequencing (WGS).

158

Table 7.2 - Sequence polymorphismsa between 16 independent B. ovis mutants that grow b without CO2 supplementation and the wild-type parent strain Mut Chr. Pos. Mutation Locusc Annot. Desc. # 01 I 805,096 C à T BOV_RS03980 NADH-quinone T188T (ACC à oxidoreductase ACT) subunit B I 1,768,982 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA1BOV (frameshift) d 02 I 1,768,982 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA1BOV (frameshift) 03 I 1,768,980 -G BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA2BOV (frameshift) 04 I 1,768,980 -G BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA2BOV (frameshift) 05 I 1,768,982 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA1BOV (frameshift) 06 I 1,768,982 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA1BOV (frameshift) 07 I 1,768,980 -G BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA2BOV (frameshift) 08 I 1,768,984 -A BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA3BOV (frameshift) 09 I 301,445 A à G BOV_RS0175 IS5/IS1182 family L224P (CTT à transposase CCT) I 301,958 A à C BOV_RS01475 sensor histidine V53G (GTT à kinase GGT) I 420,434 G à T BOV_RS02035 hypothetical protein / intergenic ß / ß hypothetical protein BOV_RS16215 I 1,725,552 C à T BOV_RS08460 branched-chain G349D (GGC à amino acid ABC GAC) transporter ATP- binding protein

159

Continuation: Table 7.2 - Sequence polymorphismsa between 16 independent B. ovis b mutants that grow without CO2 supplementation and the wild-type parent strain

I 1,768,980 -G BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA2BOV (frameshift) II 117,883 (T)8 à BOV_RS10890 HTH-type quorum intergenic (T)7 à / à sensing-dependent BOV_RS12585 transcriptional regulator VjbR II 676,505 G à A BOV_RS13635 sulfurtransferase pseudogene FdhD (415/810 nt) II 908,237 C à T BOV_RS14715 a-hydroxy-acid intergenic à / à oxidizing protein BOV_RS14720 lldD / porin family protein 10 I 805,096 C à T BOV_RS03980 NADH-quinone T188T (ACC à oxidoreductase ACT) subunit B I 1,768,982 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA1BOV (frameshift) 11 I 1,768,982 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA1BOV (frameshift) 12d I 1,712,213 G à T BOV_RS08400 polyprenyl A16S (GCC à synthetase family TCC) protein I 1,768,979 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA4BOV (frameshift) 13 I 1,768,980 -G BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA2BOV (frameshift) 14 I 1,768,980 -G BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA2BOV (frameshift) 15 I 1,768,980 -G BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA2BOV (frameshift) 16 I 1,768,982 -C BOV_RS08635 b-carbonic bcaABOV à anhydrase - bcaABOV bcaA1BOV (frameshift)

160

Continuation: Table 7.2 - Sequence polymorphismsa between 16 independent B. ovis mutants b that grow without CO2 supplementation and the wild-type parent strain Abbreviations: Mut# = mutant number, Chr .= Chromosome, Pos. = Position, Annot. = Gene Annotation, Descr. = Description. aAll mutations shown in the Table are observed in 100% of the reads from the indicated strain b We observed four differences between the B. ovis ATCC 25840 sequence deposited in GenBank and the sequence of our ATCC 25840 parent strain: In 100% of the reads, there was an insertion of a single G in BOV_RS01360 on chromosome I (NC_009505), at base 282,040; an insertion of a single C in an intergenic region on chromosome II (NC_009504), at position 20; and an insertion of a single G in an intergenic region at base 459,010, also on chromosome II. These differences were also present in all derived mutant strains. In ~56% of the sequence reads from the parental strain, we observed a C -> T transition (resulting in a T14I coding change) in BOV_RS10895, vjbR at base 118,034 of chromosome II. Each derived mutant strain carried one of several different vjbR polymorphisms, each present in 100% of the sequencing reads. cFor intergenic polymorphisms, ß or à indicate direction of flanking genes dThis strain lacks the insertion in BOV_RS01360 observed in our parental strain, and thus is identical to the deposited B. ovis ATCC 25840 GenBank sequence strain at this locus.

Table 7.3 - B. ovis ATCC 25840 Tn-Himar library Estimated number of Tn strainsa 1.8 ´ 106 Unique Tn insertions passing mapping criteria 83,660 Numbers of unique TA sites hitb 50,984 Unique Tn insertions in central 10%-90% of genes 52,612 Median Tn per protein-coding gene 15 Number of protein-coding genes hit 2,753 (82.7%) Number of unhit genesc 488 (14.4%) aChao2 estimate of the total number of barcodes in the library bSites on opposite strand count as unique cGenes > 300nt without good insertions were counted as ‘unhit’

161

Table 7.4 - B. abortus biovar association with bcaA allele clusters

b a Cl. 1 Cl. 7 Biovar CO2 dep. Tot. Cl. 2 Cl. 3 Cl. 4 Cl. 5 Cl. 6 Cl. 8 (bcaABAB) (bcaABAB2308) 196a 179 8 Biovar 1c + (–) 0 2 (1%) 0 0 7 (3%) 0 (52%) (89%) (4%) 18a 1 2 1 1 Biovar 2 + 13 (72%) 0 0 0 (5%) (5%) (11%) (5%) (5%) 14 1 2 Biovar 3 + (–) 10 (71%) 1 (7%) 0 0 0 0 (4%) (7%) (14%) 16 1 Biovar 4 + (–) 15 (94%) 0 0 0 0 0 0 (4%) (6%) 3 3 Biovar 5 – 0 0 0 0 0 0 0 (1%) (100%) 11 10 1 Biovar 6 – 0 0 0 0 0 0 (3%) (91%) (9%) 5 4 Biovar 7 – 1 (20%) 0 0 0 0 0 0 (1%) (80%) 6 4 Biovar 9 – or + 2 (33%) 0 0 0 0 0 0 (2%) (67%) 105b 23 11 4 2 2 N/A 57 (55%) 0 5 (5%) (28%) (22%) (10%) (4%) (2%) (2%)

Abbreviations: Cl. = Clutser, Tot. = Total, CO2 dep. = CO2 dependence. a Symbols indicating CO2 dependence as per Alton et al.(215): + = CO2 dependent; + (–) = mostly CO2 dependent; – = CO2 independent. bTwo strains are annotated as 2308, thus are possibly biovar 1 and redundant. Biovar was not assigned. cOne strain is annotated as “atypical”

162

Table 7.5 - Gene and pseudogene comparison across B. ovis and B. abortus ATCC 2308 strains

Polymorphisms

a Polymorphisms

between B. ovis and B.

locus within B. ovis

abortus genomes SNPs Singl

Gene Indel B. ovis B. ovis Group e nt

old locus # # new new locus s c Gene Gene length Tot Syn Indel site(s) genomes

B. abortus >1nt s d e BOV_1 BOV_R BAB_R ureF2 732 3 / 2 0 0 0 316 S06515 S22515 BOV_1 BOV_R BAB_R 1 ureT 1050 2 / 0 1 15 319 S06530 S22530 (56nt) BOV_0 BOV_R BAB_R ureG1 627 1 / 0 0 0 0 287 S01465 S17375 BOV_1 BOV_R BAB_R ureE2 606 2 / 2 0 0 0 315 S06510 S22510 BOV_0 BOV_R BAB_R 1 ureC1 1713 5 / 0 0 0 284 S01450 S17360 (30nt) BOV_2 BOV_R BAB_R pckA 1476 8 / 1 0 1 1 009 S09880 S25895 BOV_ BOV_R BAB_R eryA 1554 6 / 0 0 0 0 A0811 S14445 S28140 BOV_ BOV_R BAB_R 1 eryD 951 4 / 0 0 0

A0814 S14460 S28125 (7nt) BOV_ BOV_R BAB_R 1 eryF 948 4 / 0 0 0 A0805 S14425 S28160 (9nt) BOV_ BOV_R BAB_R 1 eryG 1041 1 / 0 0 0 A0806 S14430 S28155 (2nt) Pseudogenes BOV_ BOV_R BAB_R gluP 1239 3 / 1 0 0 0 A0172 S11215 S27245 BOV_0 BOV_R BAB_R ccoO 732 1 / 1 0 1 11 378 S01915 S17800 BOV_R BAB_R coxB / 889 2 / 1 0 0 0 S02370 S18265 2 BOV_0 BOV_R BAB_R coxG 879 2 / 1 (56nt, 0 0 478 S16275 S18290 6nt) 2 BOV_R BAB_R ctaE / 570 2 / 0 (12nt, 0 0 S11490 S30835 27nt) BOV_0 BOV_R BAB_R 1 ctaG 606 1 / 0 0 0 477 S02390 S18280 (5nt) BOV_R BAB_R 1 norB / 1350 6 / 0 1 1 S11505 S30820 (80nt)

163

Continuation: Table 7.5 - Gene and pseudogene comparison across B. ovis and B. abortus ATCC 2308 strains 1 copA/ BOV_R BAB_R / 2259 5 / 0 (153n 0 0 fixI S01885 S17775 t) BOV_R BAB_R 1 bcaA / 645 0 / 1 0 0 S08635 S24650 (2nt) BOV_0 BOV_R BAB_R ureD1 843 2 1 0 0 0 0 281 S01430 S17340 BOV_0 BOV_R BAB_R ureA1 303 1 0 0 0 0 0 202 S01435 S17345 BOV_0 BOV_R BAB_R ureB1 306 0 0 0 0 0 0 283 S01445 S17355

BOV_0 BOV_R BAB_R ureE1 516 5 3 0 0 0 0

285 S01455 S17365 BOV_0 BOV_R BAB_R ureF1 688 2 0 0 0 1 1 286 S01460 S17370 BOV_1 BOV_R BAB_R ureA2 303 1 0 0 0 0 0 312 S06495 S22495 Urease genes Urease BOV_1 BOV_R BAB_R ureB2(not pseudogenes) 480 2 0 0 0 0 0 313 S06500 S22500 BOV_1 BOV_R BAB_R ureC2 1722 5 1 0 0 0 0 314 S06505 S22505 BOV_1 BAB_R BAB_R ureG2 639 1 1 0 0 0 0 381 S22520 S22520 BOV_1 BOV_R BAB_R 1 ureD2 915 4 1 0 0 0 318 S06525 S22525 (6nt) virB1 BOV_ BOV_R BAB_R 1086 2 2 0 0 0 0 1 A0054 S10610 S26635 virB1 BOV_ BOV_R BAB_R 1 1167 3 1 0 1 1 0 A0055 S10615 S26640 (24nt)

BOV_ BOV_R BAB_R virB9 870 0 0 0 0 1 1 A0056 S10620 S26645 BOV_ BOV_R BAB_R virB8 720 2 1 0 0 0 0 A0057 S10625 S26650 BOV_R BAB_R virB7 / 174 0 0 0 0 0 0 S10630 S26655 BOV_ BOV_R BAB_R virB6 1044 1 1 0 0 1 11 A0058 S10635 S26660 BOV_ BOV_R BAB_R virB5 717 4 2 0 0 0 0 A0059 S10640 S26665 Type Secretion IV System BOV_ BOV_R BAB_R virB4 2496 4 1 0 0 2 2f A0060 S10645 S26670 BOV_ BOV_R BAB_R virB3 351 1 1 1 0 0 0 A0061 S10650 S26675

164

Continuation: Table 7.5 - Gene and pseudogene comparison across B. ovis and B. abortus ATCC 2308 strains BOV_ BOV_R BAB_R virB2 318 0 0 0 0 0 0 A0062 S10655 S26680

BOV_ BOV_R BAB_R virB1 717 2 0 0 0 0 0 A0063 S10660 S26685 aIn B. abortus ATCC 2308 bNumber of polymorphisms between all 17 sequenced B. ovis clinical isolates (see Data Sheet 1) and B. abortus ATCC 2308 strain (GenBank accessions NC_007618 and NC_007624). cSyn = synonymous mutations. In the case of pseudogenes, synonymous mutations were not noted. dNumber of polymorphic sites among sequences from B. ovis isolates eNumber of B. ovis sequences that harbor the polymorphism fTwo genomes have distinct polymorphisms

Table 7.6

Number of cells Cells in each Timepoint OD Volume 600 collected PCR rxn

T0 0.6 100 ml 4 ´ 109 8 ´ 107 T1 0.05 1 ml 6.6 ´ 109 1.3 ´ 108 T2 0.12 1 ml 8 ´ 109 1.6 ´ 108 T3 0.9 0.5 ml 6 ´ 109 1.2 ´ 108 T4 2.4 0.5 ml 8 ´ 109 1.6 ´ 108 rxn = reaction;

165

Table 7.7 - Genome characteristics of select Erythrobacter spp. isolates

–

% GC orphan) LOV domain Genome Genome status Phototrophy genes

Source Genome length (Mb) LOV kinases LOV(SDRR kinases Isolate a (reference) HWE / kinases HisKA2 E. longus Seaweed in WGS; 14 contigs 3.6 57 YES 0 0 2 DSM 6997 Aburatsubo bay, (298) (Och 101) Kanagawa, Japan (343) E. litoralis Cyanobacterial WGS; 22 contigs 3.21 65 YES 1 1-0 5 DSM 8509 mat, Supralitoral (298) zone, Texal, -- Netherlands (301, Complete (this 302) reference) Erythrobacter Surface sediment, WGS; 19 contigs 2.97 63 NO 1 0-1 16 sp. SD-21 San Diego Bay, USA (344) Erythrobacter Surface water, WGS; 4 contigs 3.26 61 YES 1 0-1 6 sp. NAP1 Northwest (346) Atlantic Ocean (345) E. litoralis 10 m deep Complete (297) 3.05 63 NO 4 1-2 8 HTCC2594 Sargasso Sea, Atlantic Ocean (297) a Isolates sequenced at the time this work was initiated b Extracted from Glantz et al., 2016 (275) c Subset of genes identified in Glantz et al., 2016 (275) with kinase domains in an operon with a single domain response regulator (SDRR) or encoded as an orphan gene. All of these are HWE/HisKA_2-family kinases. d Identified in the MIST database 2.0 (307)

166

8. REFERENCES

1. Kolter R, Siegele DA, Tormo A. 1993. The stationary phase of the bacterial life cycle. Annu Rev Microbiol 47:855–874. 2. Huisman G, Siegele DA, Zambrano MM, Kolter R. 1996. Morphological and Physiological Changes During Stationary Phase. 3. Finkel SE. 2006. Long-term survival during stationary phase: evolution and the GASP phenotype. Nat Rev Microbiol 4:113–120. 4. Lange R, Hengge-Aronis R. 1991. Growth phase-regulated expression of bolA and morphology of stationary-phase Escherichia coli cells are controlled by the novel sigma factor sigma S. J Bacteriol 173:4474–4481. 5. Jenkins DE, Schultz JE, Matin A. 1988. Starvation-induced cross protection against heat or H2O2 challenge in Escherichia coli. J Bacteriol 170:3910–3914. 6. Jenkins DE, Chaisson SA, Matin A. 1990. Starvation-induced cross protection against osmotic challenge in Escherichia coli. J Bacteriol 172:2779–2781. 7. Rinas U, Hellmutii K, Kang R, Seeger A, Schlieker H. 1995. Entry of Escherichia coli into stationary phase is indicated by endogenous and exogenous accumulation of nucleobases. Appl Environ Microbiol 61:4147–4151. 8. Lange R, Hengge-Aronis R. 1991. Identification of a central regulator of stationary-phase gene expression in Escherichia coli. Mol Microbiol 5:49–59. 9. Gentry DR, Hernandez VJ, Nguyen LH, Jensen DB, Cashel M. 1993. Synthesis of the stationary-phase sigma factor σ(s) is positively regulated by ppGpp. J Bacteriol 175:7982– 7989. 10. Schellhorn HE. 2020. Function, Evolution, and Composition of the RpoS Regulon in Escherichia coli. Front Microbiol 11:560099. 11. Battesti A, Majdalani N, Gottesman S. 2011. The RpoS-mediated general stress response in Escherichia coli. Annu Rev Microbiol2011/06/07. 65:189–213. 12. Loewen PC, Hengge-Aronis R. 1994. The role of the sigma factor sigma S (KatF) in bacterial global regulation. Annu Rev Microbiol 48:53–80. 13. Gottesman S. 2019. Trouble is coming: Signaling pathways that regulate general stress responses in bacteria. J Biol Chem 294:11685–11700. 14. Galperin MY. 2018. What bacteria want. Environ Microbiol 20:4221–4229. 15. Franze de Fernandez MT, Hayward WS, August JT. 1972. Bacterial proteins required for replication of phage Q ribonucleic acid. Pruification and properties of host factor I, a ribonucleic acid-binding protein. J Biol Chem 247:824–831. 16. Muffler A, Fischer D, Hengge-Aronis R. 1996. The RNA-binding protein HF-I, known as a host factor for phage Qbeta RNA replication, is essential for rpoS translation in

167

Escherichia coli. Genes Dev 10:1143–1151. 17. Nogueira T, Springer M. 2000. Post-transcriptional control by global regulators of gene expression in bacteria. Curr Opin Microbiol 3:154–158. 18. Hengge-Aronis R. 1996. Back to log phase: σS as a global regulator in the osmotic control of gene expression in Escherichia coli. Mol Microbiol 21:887–893. 19. Milewska K, Krause K, Szalewska-Pałasz A. 2020. The stringent response of marine bacteria – assessment of (p)ppGpp accumulation upon stress conditions. J Appl Genet 61:123–130. 20. Pacios O, Blasco L, Bleriot I, Fernandez-Garcia L, Ambroa A, López M, Bou G, Cantón R, Garcia-Contreras R, Wood TK, Tomás M. 2020. (p)ppGpp and Its Role in Bacterial Persistence: New Challenges. Antimicrob Agents Chemother 64:e01283-20. 21. Okada Y, Makino S, Tobe T, Okada N, Yamazaki S. 2002. Cloning of rel from Listeria monocytogenes as an Osmotolerance Involvement Gene. Appl Environ Microbiol 68:1541 LP – 1547. 22. Cashel M, Gallant J. 1969. Two compounds implicated in the function of the RC gene of Escherichia coli. Nature 221:838–841. 23. Traxler MF, Summers SM, Nguyen HT, Zacharia VM, Hightower GA, Smith JT, Conway T. 2008. The global, ppGpp-mediated stringent response to amino acid starvation in Escherichia coli. Mol Microbiol 68:1128–1148. 24. Chatterji D, Ojha AK. 2001. Revisiting the stringent response, ppGpp and starvation signaling. Curr Opin Microbiol 4:160–165. 25. Dozot M, Boigegrain RA, Delrue RM, Hallez R, Ouahrani-Bettache S, Danese I, Letesson JJ, De Bolle X, Köhler S. 2006. The stringent response mediator Rsh is required for Brucella melitensis and Brucella suis virulence, and for expression of the type IV secretion system virB. Cell Microbiol 8:1791–1802. 26. Kim S, Watanabe K, Suzuki H, Watarai M. 2005. Roles of Brucella abortus SpoT in morphological differentiation and intramacrophagic replication. Microbiology 151:1607– 1617. 27. Nazir A, Harinarayanan R. 2016. (p)ppGpp and the bacterial cell cycle. J Biosci 41:277– 282. 28. Choy HE. 2000. The study of guanosine 5’-diphosphate 3’-diphosphate-mediated transcription regulation in vitro using a coupled transcription-translation system. J Biol Chem 275:6783–6789. 29. Durfee T, Hansen AM, Zhi H, Blattner FR, Ding JJ. 2008. Transcription profiling of the stringent response in Escherichia coli. J Bacteriol 190:1084–1096. 30. Boutte CC, Crosson S. 2011. The complex logic of stringent response regulation in Caulobacter crescentus: Starvation signalling in an oligotrophic environment. Mol Microbiol 80:695–714. 31. Lesley JA, Shapiro L. 2008. SpoT regulates DnaA stability and initiation of DNA replication in carbon-starved Caulobacter crescentus. J Bacteriol 190:6867–6880.

168

32. Chiaverotti TA, Parker G, Gallant J, Agabian N. 1981. Conditions that trigger guanosine tetraphosphate accumulation in Caulobacter crescentus. J Bacteriol 145:1463–1465. 33. Geiger T, Kästle B, Gratani FL, Goerke C, Wolz C. 2014. Two small (p)ppGpp synthases in Staphylococcus aureus mediate tolerance against cell envelope stress conditions. J Bacteriol 196:894–902. 34. Corrigan RM, Bellows LE, Wood A, Gründling A. 2016. ppGpp negatively impacts ribosome assembly affecting growth and antimicrobial tolerance in Gram-positive bacteria. Proc Natl Acad Sci U S A 113:E1710-9. 35. Hecker M, Pane-Farre J, Volker U. 2007. SigB-dependent general stress response in Bacillus subtilis and related gram-positive bacteria. Annu Rev Microbiol2007/11/24. 61:215–236. 36. Benson AK, Haldenwang WG. 1993. Regulation of sigma B levels and activity in Bacillus subtilis. J Bacteriol 175:2347–2356. 37. Völker U, Engelmann S, Maul B, Riethdorf S, Völker A, Schmid R, Mach H, Hecker M. 1994. Analysis of the induction of general stress proteins of Bacillus subtilis. Microbiology 140 ( Pt 4:741–752. 38. Yang X, Kang CM, Brody MS, Price CW. 1996. Opposing pairs of serine protein kinases and phosphatases transmit signals of environmental stress to activate a bacterial transcription factor. Genes Dev 10:2265–2275. 39. Britos L, Abeliuk E, Taverner T, Lipton M, McAdams H, Shapiro L. 2011. Regulatory response to carbon starvation in Caulobacter crescentus. PLoS One2011/04/16. 6:e18179. 40. Fiebig A, Herrou J, Willett J, Crosson S. 2015. General Stress Signaling in the Alphaproteobacteria. Annu Rev Genet 49:603–625. 41. Staron A, Mascher T. 2010. General stress response in alpha-proteobacteria: PhyR and beyond. Mol Microbiol 78:271–277. 42. Francez-Charlot A, Kaczmarczyk A, Fischer HM, Vorholt JA. 2015. The general stress response in Alphaproteobacteria. Trends Microbiol 23:164–171. 43. Poindexter JS. 1964. Biological properties and classification of teh Caulobacter group. Bacteriol Rev 28:231–295. 44. Orcutt BN, Sylvan JB, Knab NJ, Edwards KJ. 2011. Microbial ecology of the dark ocean above, at, and below the seafloor. Microbiol Mol Biol Rev 75:361–422. 45. Portillo MC, Gonzalez JM. 2008. Microbial communities and immigration in volcanic environments of Canary Islands (Spain). Naturwissenschaften 95:307–315. 46. Giovannoni SJ, Tripp HJ, Givan S, Podar M, Vergin KL, Baptista D, Bibbs L, Eads J, Richardson TH, Noordewier M, Rappé MS, Short JM, Carrington JC, Mathur EJ. 2005. Genome Streamlining in a Cosmopolitan Oceanic Bacterium. Science (80- ) 309:1242 LP – 1245. 47. Kaneko T, Nakamura Y, Sato S, Asamizu E, Kato T, Sasamoto S, Watanabe A, Idesawa K, Ishikawa A, Kawashima K, Kimura T, Kishida Y, Kiyokawa C, Kohara M, Matsumoto M, Matsuno A, Mochizuki Y, Nakayama S, Nakazaki N, Shimpo S, Sugimoto M, Takeuchi C,

169

Yamada M, Tabata S. 2000. Complete Genome Structure of the Nitrogen-fixing Symbiotic Bacterium Mesorhizobium loti. DNA Res 7:331–338. 48. Li H, Tong Y, Huang Y, Bai J, Yang H, Liu W, Cao W. 2012. Complete genome sequence of Bartonella quintana, a bacterium isolated from rhesus macaques. J Bacteriol 194:6347. 49. Ettema TJG, Andersson SGE. 2009. The alpha-proteobacteria: the Darwin finches of the bacterial world. Biol Lett 5:429–432. 50. Philippot L, Andersson SGEE, Battin TJ, Prosser JI, Schimel JP, Whitman WB, Hallin S. 2010. The ecological coherence of high bacterial taxonomic ranks. Nat Rev Microbiol 8:523–529. 51. Batut J, Andersson SG, O’Callaghan D. 2004. The evolution of chronic infection strategies in the alpha-proteobacteria. Nat Rev Microbiol2004/11/20. 2:933–945. 52. Bruce D. 1887. Note on the discovery of a microorganism in Malta Fever. [John Brigg], [London]. 53. Bang B. 1897. Die Aetiologie des seuchenhaften (“infectiösen”) Verwerfens. Zietschrift für tiermedizin 1:241–278. 54. Seleem MN, Boyle SM, Sriranganathan N. 2010. Brucellosis: A re-emerging zoonosis. Vet Microbiol 140:392–398. 55. Boschiroli ML, Foulongne V, O’Callaghan D. 2001. Brucellosis: A worldwide zoonosis. Curr Opin Microbiol 4:58–64. 56. Galińska EM, Zagórski J. 2013. Brucellosis in humans - Etiology, diagnostics, clinical forms. Ann Agric Environ Med 20:233–238. 57. Ridler AL, West DM. 2011. Control of Brucella ovis Infection in Sheep. Vet Clin North Am - Food Anim Pract 27:61–66. 58. Godfroid J, Garin-Bastuji B, Saegerman C, Blasco JM. 2013. Brucellosis in terrestrial wildlife. OIE Rev Sci Tech 32:27–42. 59. Moreno E. 2014. Retrospective and prospective perspectives on zoonotic brucellosis. Front Microbiol 5:1–18. 60. Whatmore AM. 2009. Current understanding of the genetic diversity of Brucella, an expanding genus of zoonotic pathogens. Infect Genet Evol 9:1168–1184. 61. Alton GG, Forsy JRL. 1996. Brucella, p. . In Baron, S (ed.), Medical Microbiology. Galveston (TX). 62. Franco MP, Mulder M, Gilman RH, Smits HL. 2007. Human brucellosis. Lancet Infect Dis 7:775–786. 63. Al Dahouk S, Hofer E, Tomaso H, Vergnaud G, Le Flèche P, Cloeckaert A, Koylass MS, Whatmore AM, Nöckler K, Scholz HC. 2012. Intraspecies diodiversity of the genetically homologous species Brucella microti. Appl Environ Microbiol 78:1534–1543. 64. Spink WW, Morisset R. 1970. Epidemic canine brucellosis due to a new species: Brucella canis. Trans Am Clin Climatol Assoc 81:43–50. 65. Buddle M, Boyes B. 1953. A Brucella mutant causing genital disease of sheep in New 170

Zealand. Aust Vet J 29:145–153. 66. STOENNER HG, LACKMAN DB. 1957. A new species of Brucella isolated from the desert wood rat, Neotoma lepida Thomas. Am J Vet Res 18:947–951. 67. Foster G, Jahans KL, Reid RJ, Ross HM. 1996. Isolation of Brucella species from cetaceans, seals and an otter. Vet Rec 138:583–586. 68. Ross HM, Foster G, Reid RJ, Jahans KL, MacMillan AP. 1994. Brucella species infection in sea-mammals. Vet Rec. England. 69. Ewalt DR, Payeur JB, Martin BM, Cummins DR, Miller WG. 1994. Characteristics of a Brucella species from a bottlenose dolphin (Tursiops truncatus). J Vet diagnostic Investig Off Publ Am Assoc Vet Lab Diagnosticians, Inc 6:448–452. 70. Hubálek Z, Scholz HC, Sedlácek I, Melzer F, Sanogo YO, Nesvadbová J. 2007. Brucellosis of the common vole (Microtus arvalis). Vector Borne Zoonotic Dis 7:679–687. 71. Scholz HC, Nöckler K, Göllner C, Bahn P, Vergnaud G, Tomaso H, Al Dahouk S, Kämpfer P, Cloeckaert A, Maquart M, Zygmunt MS, Whatmore AM, Pfeffer M, Huber B, Busse H- J, De BK. 2010. Brucella inopinata sp. nov., isolated from a breast implant infection. Int J Syst Evol Microbiol 60:801–808. 72. Whatmore AM, Davison N, Cloeckaert A, Al Dahouk S, Zygmunt MS, Brew SD, Perrett LL, Koylass MS, Vergnaud G, Quance C, Scholz HC, Dick EJ, Hubbard G, Schlabritz- Loutsevitch NE. 2014. Brucella papionis sp. nov., isolated from baboons (Papio spp.). Int J Syst Evol Microbiol 64:4120–4128. 73. Scholz HC, Revilla-Fernández S, Dahouk S Al, Hammerl JA, Zygmunt MS, Cloeckaert A, Koylass M, Whatmore AM, Blom J, Vergnaud G, Witte A, Aistleitner K, Hofer E. 2016. Brucella vulpis sp. nov., isolated from mandibular lymph nodes of red foxes (Vulpes vulpes). Int J Syst Evol Microbiol 66:2090–2098. 74. Soler-Lloréns PF, Quance CR, Lawhon SD, Stuber TP, Edwards JF, Ficht TA, Robbe- Austerman S, O’Callaghan D, Keriel A. 2016. A Brucella spp. isolate from a Pac-Man frog (Ceratophrys ornata) reveals characteristics departing from classical brucellae. Front Cell Infect Microbiol 6:1–16. 75. Al Dahouk S, Köhler S, Occhialini A, Jiménez de Bagüés MP, Hammerl JA, Eisenberg T, Vergnaud G, Cloeckaert A, Zygmunt MS, Whatmore AM, Melzer F, Drees KP, Foster JT, Wattam AR, Scholz HC. 2017. Brucella spp. of amphibians comprise genomically diverse motile strains competent for replication in macrophages and survival in mammalian hosts. Sci Rep 7:44420. 76. Fiebig A, Vrentas CE, Le T, Huebner M, Boggiatto PM, Olsen SC, Crosson S. 2021. Quantification of Brucella abortus population structure in a natural host. Proc Natl Acad Sci U S A 118. 77. Atluri VL, Xavier MN, de Jong MF, den Hartigh AB, Tsolis RM. 2011. Interactions of the human pathogenic Brucella species with their hosts. Annu Rev Microbiol 65:523–541. 78. de Figueiredo P, Ficht TA, Rice-Ficht A, Rossetti CA, Adams LG. 2015. Pathogenesis and immunobiology of brucellosis: review of Brucella-host interactions. Am J Pathol 185:1505– 1517.

171

79. Tuon FF, Gondolfo RB, Cerchiari N. 2017. Human-to-human transmission of Brucella – a systematic review. Trop Med Int Heal 22:539–546. 80. Tsolis RM, Seshadri R, Santos RL, Sangari FJ, García Lobo JM, de Jong MF, Ren Q, Myers G, Brinkac LM, Nelson WC, DeBoy RT, Angiuoli S, Khouri H, Dimitrov G, Robinson JR, Mulligan S, Walker RL, Elzer PE, Hassan KA, Paulsen IT. 2009. Genome degradation in Brucella ovis corresponds with narrowing of its host range and tissue tropism. PLoS One2018/05/23. 4:https://doi:10.1371/journal.pone.0005519. 81. De Jong MF, Rolán HG, Tsolis RM. 2010. Microreview: Innate immune encounters of the (Type) 4th kind: Brucella. Cell Microbiol 12:1195–1202. 82. Tsolis RM, Young GM, Solnick J V., Bäumler AJ. 2008. From bench to bedside: Stealth of enteroinvasive pathogens. Nat Rev Microbiol 6:883–892. 83. Ahmed W, Zheng K, Liu ZF. 2016. Establishment of chronic infection: Brucella’s stealth strategy. Front Cell Infect Microbiol 6:1–12. 84. Teng TS, Ji AL, Ji XY, Li YZ. 2017. Neutrophils and immunity: From bactericidal action to being conquered. J Immunol Res 2017. 85. Mora-Cartín R, Gutiérrez-Jiménez C, Alfaro-Alarcón A, Chaves-Olarte E, Chacón-Díaz C, Barquero-Calvo E, Moreno E. 2019. Neutrophils Dampen Adaptive Immunity in Brucellosis. Infect Immun 87. 86. Ben-Tekaya H, Gorvel J-P, Dehio C. 2013. Bartonella and Brucella--weapons and strategies for stealth attack. Cold Spring Harb Perspect Med 3. 87. Lu YC, Yeh WC, Ohashi PS. 2008. LPS/TLR4 signal transduction pathway. Cytokine 42:145–151. 88. Conde-Alvarez R, Arce-Gorvel V, Iriarte M, Mancek-Keber M, Barquero-Calvo E, Palacios-Chaves L, Chacon-Diaz C, Chaves-Olarte E, Martirosyan A, von Bargen K, Grillo MJ, Jerala R, Brandenburg K, Llobet E, Bengoechea JA, Moreno E, Moriyon I, Gorvel JP. 2012. The lipopolysaccharide core of Brucella abortus acts as a shield against innate immunity recognition. PLoS Pathog 8:e1002675. 89. Weiss DS, Takeda K, Akira S, Zychlinsky A, Moreno E. 2005. MyD88, but not toll-like receptors 4 and 2, is required for efficient clearance of Brucella abortus. Infect Immun 73:5137–5143. 90. Kawasaki T, Kawai T. 2014. Toll-Like Receptor Signaling Pathways . Front Immunol . 91. Andersen-Nissen E, Smith KD, Strobe KL, Barrett SLR, Cookson BT, Logan SM, Aderem A. 2005. Evasion of Toll-like receptor 5 by flagellated bacteria. Proc Natl Acad Sci U S A 102:9247–9252. 92. Letesson JJ, Lestrate P, Delrue RM, Danese I, Bellefontaine F, Fretin D, Taminiau B, Tibor A, Dricot A, Deschamps C, Haine V, Leonard S, Laurent T, Mertens P, Vandenhaute J, De Bolle X. 2002. Fun stories about Brucella: The “furtive nasty bug.” Vet Microbiol 90:317– 328. 93. Fretin D, Fauconnier A, Köhler S, Halling S, Léonard S, Nijskens C, Ferooz J, Lestrate P, Delrue RM, Danese I, Vandenhaute J, Tibor A, DeBolle X, Letesson JJ. 2005. The sheathed

172

flagellum of Brucella melitensis is involved in persistence in a murine model of infection. Cell Microbiol 7:687–698. 94. Covert J, Mathison AJ, Eskra L, Banai M, Splitter G. 2009. Brucella melitensis, B. neotomae and B. ovis elicit common and distinctive macrophage defense transcriptional responses. Exp Biol Med 234:1450–1467. 95. Celli J, De Chastellier C, Franchini DM, Pizarro-Cerda J, Moreno E, Gorvel JP. 2003. Brucella evades macrophage killing via VirB-dependent sustained interactions with the endoplasmic reticulum. J Exp Med 198:545–556. 96. Celli J. 2015. The changing nature of the Brucella-containing vacuole. Cell Microbiol 17:951–958. 97. Celli J. 2019. The Intracellular Life Cycle of Brucella spp. Bact Intracellularity 7:101–111. 98. Sá JC, Silva TMA, Costa ÉA, Silva APC, Tsolis RM, Paixão TA, Carvalho Neta A V., Santos RL. 2012. The virB-encoded type IV secretion system is critical for establishment of infection and persistence of Brucella ovis infection in mice. Vet Microbiol 159:130–140. 99. Delrue RM, Martinez-Lorenzo M, Lestrate P, Danese I, Bielarz V, Mertens P, De Bolle X, Tibor A, Gorvel JP, Letesson JJ. 2001. Identification of Brucella spp. genes involved in intracellular trafficking. Cell Microbiol 3:487–497. 100. Starr T, Child R, Wehrly TD, Hansen B, Hwang S, López-Otin C, Virgin HW, Celli J. 2012. Selective subversion of autophagy complexes facilitates completion of the Brucella intracellular cycle. Cell Host Microbe 11:33–45. 101. Roop RM, Gaines JM, Anderson ES, Caswell CC, Martin DW. 2009. Survival of the fittest: How Brucella strains adapt to their intracellular niche in the host. Med Microbiol Immunol 198:221–238. 102. Roop II RM, Gee JM, Robertson GT, Richardson JM, Ng W-L, Winkler ME. 2003. Brucella stationary-phase gene expression and virulence. Annu Rev Microbiol 57:57–76. 103. Jiang X, Leonard B, Benson R, Baldwin CL. 1993. Macrophage control of Brucella abortus: Role of reactive oxygen intermediates and nitric oxide. Cell Immunol 151:309–319. 104. Muraille E, Leo O, Moser M. 2014. Th1/Th2 paradigm extended: Macrophage polarization as an unappreciated pathogen-driven escape mechanism? Front Immunol 5:1–12. 105. Brundu S FA. 2015. Polarization and repolarization of macrophages. J Clin Cell Immunol 06:1–10. 106. Reniere ML. 2018. Reduce, Induce, Thrive: Bacterial Redox Sensing during Pathogenesis. J Bacteriol 200:1–27. 107. Robertson GT, Roop RM. 1999. The Brucella abortus host factor I (HF-I) protein contributes to stress resistance during stationary phase and is a major determinant of virulence in mice. Mol Microbiol 34:690–700. 108. Hanna N, Ouahrani-Bettache S, Drake KL, Adams LG, Köhler S, Occhialini A. 2013. Global Rsh-dependent transcription profile of Brucella suis during stringent response unravels adaptation to nutrient starvation and cross-talk with other stress responses. BMC Genomics 14:459.

173

109. Yang P, Huang S, Yan X, Huang G, Dong X, Zheng T, Yuan D, Wang R, Li R, Tan Y, Xu A. 2014. Origin of the phagocytic respiratory burst and its role in gut epithelial phagocytosis in a basal chordate. Free Radic Biol Med 70:54–67. 110. Halliwell B, Gutteridge JM. 1984. Oxygen toxicity, oxygen radicals, transition metals and disease. Biochem J 219:1–14. 111. Ezraty B, Gennaris A, Barras F, Collet JF. 2017. Oxidative stress, protein damage and repair in bacteria. Nat Rev Microbiol 15:385–396. 112. Phaniendra A, Jestadi DB, Periyasamy L. 2015. Free radicals: properties, sources, targets, and their implication in various diseases. Indian J Clin Biochem 30:11–26. 113. Imlay JA. 2013. The molecular mechanisms and physiological consequences of oxidative stress: Lessons from a model bacterium. Nat Rev Microbiol 11:443–454. 114. McCord JM, Fridovich I. 1969. Superoxide dismutase. An enzymic function for erythrocuprein (hemocuprein). J Biol Chem 244:6049–6055. 115. Aust SD, Morehouse LA, Thomas CE. 1985. Role of metals in oxygen radical reactions. J Free Radic Biol Med 1:3–25. 116. Chelikani P, Fita I, Loewen PC. 2004. Diversity of structures and properties among catalases. Cell Mol Life Sci 61:192–208. 117. Brigelius-Flohé R, Maiorino M. 2013. Glutathione peroxidases. Biochim Biophys Acta 1830:3289–3303. 118. Rhee SG, Kang SW, Chang TS, Jeong W, Kim K. 2001. Peroxiredoxin, a novel family of peroxidases. IUBMB Life 52:35–41. 119. Forrester MT, Foster MW. 2012. Protection from nitrosative stress: A central role for microbial flavohemoglobin. Free Radic Biol Med 52:1620–1633. 120. Gardner AM, Helmick RA, Gardner PR. 2002. Flavorubredoxin, an inducible catalyst for nitric oxide reduction and detoxification in Escherichia coli. J Biol Chem 277:8172–8177. 121. Winterbourn CC, Hampton MB. 2008. Thiol chemistry and specificity in redox signaling. Free Radic Biol Med 45:549–561. 122. Vogt W. 1995. Oxidation of methionyl residues in proteins: tools, targets, and reversal. Free Radic Biol Med 18:93–105. 123. Meister A, Anderson ME. 1983. Glutathione. Annu Rev Biochem 52:711–760. 124. Masip Ll, Veeravalli K, Georgiou G. 2006. The many faces of glutathione in bacteria. Antioxidants Redox Signal 8:753–762. 125. Kerksick C, Willoughby D. 2005. The antioxidant role of glutathione and N-acetyl-cysteine supplements and exercise-induced oxidative stress. J Int Soc Sports Nutr 2:38–44. 126. Park S, Imlay JA. 2003. High levels of intracellular cysteine promote oxidative DNA damage by driving the Fenton reaction. J Bacteriol 185:1942–1950. 127. Reddy VN. 1990. Glutathione and its function in the lens--an overview. Exp Eye Res 50:771–778.

174

128. Fahey RC. 2013. Glutathione analogs in prokaryotes. Biochim Biophys Acta 1830:3182– 3198. 129. Winterbourn CC, Kettle AJ. 2013. Redox reactions and microbial killing in the neutrophil phagosome. Antioxid Redox Signal 18:642–660. 130. Spitznagel JK. 1977. Bactericidal mechanisms of the granulocyte. Prog Clin Biol Res 13:103–131. 131. Fang FC. 2004. Antimicrobial reactive oxygen and nitrogen species: Concepts and controversies. Nat Rev Microbiol 2:820–832. 132. Storz G, Tartaglia LA. 1992. OxyR: a regulator of antioxidant genes. J Nutr 122:627–630. 133. Dubbs JM, Mongkolsuk S. 2012. Peroxide-Sensing Transcriptional Regulators in Bacteria. J Bacteriol 194:5495 LP – 5503. 134. Duarte V, Latour J-M. 2010. PerR vs OhrR: selective peroxide sensing in Bacillus subtilis. Mol Biosyst 6:316–323. 135. Fritsch VN, Loi V Van, Busche T, Tung QN, Lill R, Horvatek P, Wolz C, Kalinowski J, Antelmann H. 2020. The alarmone (p)ppGpp confers tolerance to oxidative stress during the stationary phase by maintenance of redox and iron homeostasis in Staphylococcus aureus. Free Radic Biol Med 161:351–364. 136. Gee JM, Valderas MW, Kovach ME, Grippe VK, Robertson GT, Ng W-L, Richardson JM, Winkler ME, Roop 2nd RM. 2005. The Brucella abortus Cu,Zn superoxide dismutase is required for optimal resistance to oxidative killing by murine macrophages and wild-type virulence in experimentally infected mice. Infect Immun 73:2873–2880. 137. Martin DW, Baumgartner JE, Gee JM, Anderson ES, Martin Roop I. 2012. SodA is a major metabolic antioxidant in Brucella abortus 2308 that plays a significant, but limited, role in the virulence of this strain in the mouse model. Microbiol (United Kingdom) 158:1767– 1774. 138. Lensmire JM, Hammer ND. 2019. Nutrient sulfur acquisition strategies employed by bacterial pathogens. Curr Opin Microbiol 47:52–58. 139. Spyrakis F, Singh R, Cozzini P, Campanini B, Salsi E, Felici P, Raboni S, Benedetti P, Cruciani G, Kellogg GE, Cook PF, Mozzarelli A. 2013. Isozyme-specific ligands for O- acetylserine sulfhydrylase, a novel antibiotic target. PLoS One 8:e77558. 140. Schnell R, Sriram D, Schneider G. 2015. Pyridoxal-phosphate dependent mycobacterial cysteine synthases: Structure, mechanism and potential as drug targets. Biochim Biophys Acta - Proteins Proteomics 1854:1175–1183. 141. Gebhardt MJ, Gallagher LA, Jacobson RK, Usacheva EA, Peterson LR, Zurawski D V, Shuman HA. 2015. Joint Transcriptional Control of Virulence and Resistance to Antibiotic and Environmental Stress in Acinetobacter baumannii. MBio 6:e01660-15. 142. Gebhardt MJ, Czyz DM, Singh S, Zurawski D V, Becker L, Shuman HA. 2020. GigC, a LysR Family Transcription Regulator, Is Required for Cysteine Metabolism and Virulence in Acinetobacter baumannii Infect Immun 89:e00180-20.

175

143. Lithgow JK, Hayhurst EJ, Cohen G, Aharonowitz Y, Foster SJ. 2004. Role of a Cysteine Synthase in Staphylococcus aur. J Bacteriol 186:1579 LP – 1590. 144. Lensmire JM, Dodson JP, Hsueh BY, Wischer MR, Delekta PC, Shook JC, Ottosen EN, Kies PJ, Ravi J, Hammer ND. 2020. The Staphylococcus aureus Cystine Transporters TcyABC and TcyP Facilitate Nutrient Sulfur Acquisition during Infection. Infect Immun 88. 145. Connolly JP, Comerci D, Alefantis TG, Walz A, Quan M, Chafin R, Grewal P, Mujer C V, Ugalde RA, DelVecchio VG. 2006. Proteomic analysis of Brucella abortus cell envelope and identification of immunogenic candidate proteins for vaccine development. Proteomics 6:3767–3780. 146. Jain S, Afley P, Kumar S. 2013. Immunological responses to recombinant cysteine synthase A of Brucella abortus in BALB/c mice. World J Microbiol Biotechnol 29:907–913. 147. On S, Ovis B, Disease G, Sheep OF, New IN, Buddle BYMB. 1955. Genital Disease of Sheep in New Zealand. 148. Poltorak A, Smirnova I, He X, Liu MY, Van Huffel C, McNally O, Birdwell D, Alejos E, Silva M, Du X, Thompson P, Chan EK, Ledesma J, Roe B, Clifton S, Vogel SN, Beutler B. 1998. Genetic and physical mapping of the Lps locus: identification of the toll-4 receptor as a candidate gene in the critical region. Blood Cells Mol Dis 24:340–355. 149. Murdock JL, Núñez G. 2016. TLR4: The Winding Road to the Discovery of the LPS Receptor. J Immunol 197:2561–2562. 150. Liu Y, Sun J, Peng X, Dong H, Qin Y, Shen Q, Jiang H, Xu G, Feng Y, Sun S, Ding J, Chen R. 2020. Deletion of the LuxR-type regulator VjbR in Brucella canis affects expression of type IV secretion system and bacterial virulence, and the mutant strain confers protection against Brucella canis challenge in mice. Microb Pathog 139:103865. 151. Li P, Tian M, Bao Y, Hu H, Liu J, Yin Y, Ding C, Wang S, Yu S. 2017. Brucella rough mutant induce macrophage death via activating IRE1a pathway of endoplasmic reticulum stress by enhanced T4SS secretion. Front Cell Infect Microbiol 7:1–15. 152. Pei J, Ficht TA. 2004. Brucella abortus rough mutants are cytopathic for macrophages in culture. Infect Immun 72:440–450. 153. Herrou J, Willett JW, Fiebig A, Czyż DM, Cheng JX, Ultee E, Briegel A, Bigelow L, Babnigg G, Kim Y, Crosson S. 2019. Brucella periplasmic protein EipB is a molecular determinant of cell envelope integrity and virulence. bioRxiv 201:1–20. 154. CFSPH. 2007. Ovine Epididymitis : Brucella ovis. CFSPH Publ 1–4. 155. Antunes JMA de P, Allendorf SD, Appolinário CM, Cagnini DQ, Figueiredo PR, Júnior JB, Baños JV, Kocan KM, de la Fuente J, Megid J. 2013. Rough virulent strain of Brucella ovis induces pro- and anti-inflammatory cytokines in reproductive tissues in experimentally infected rams. Vet Microbiol 161:339–343. 156. Gouletsou PG, Fthenakis GC. 2015. Microbial diseases of the genital system of rams or bucks. Vet Microbiol 181:130–135. 157. Picard-Hagen N, Berthelot X, Champion JL, Eon L, Lyazrhi F, Marois M, Peglion M,

176

Schuster A, Trouche C, Garin-Bastuji B. 2015. Contagious epididymitis due to Brucella ovis: Relationship between sexual function, serology and bacterial shedding in semen. BMC Vet Res 11:1–7. 158. Alton GG, Jones LM, Pietz DE. 1975. Laboratory techniques in brucellosis. Monogr Ser World Heal Organ 1–163. 159. Meyer ME. 1969. Phenotypic comparison of Brucella ovis to the DNA-homologous Brucella species. Am J Vet Res 30:1757–1764. 160. Petersen E, Rajashekara G, Sanakkayala N, Eskra L, Harms J, Splitter G. 2013. Erythritol triggers expression of virulence traits in Brucella melitensis. Microbes Infect2013/02/16. 15:440–449. 161. Varesio LM, Willett JW, Fiebig A, Crosson S. 2019. A carbonic anhydrase pseudogene sensitizes select Brucella lineages to low CO2 tension. J Bacteriol https://doi.org/10.1128/JB.00509-19. 162. Pérez-Etayo L, De Miguel MJ, Conde-Álvarez R, Muñoz PM, Khames M, Iriarte M, Moriyón I, Zúñiga-Ripa A. 2018. The CO 2 -dependence of Brucella ovis and Brucella abortus biovars is caused by defective carbonic anhydrases. Vet Res 49:1–12. 163. Rambow-Larsen AA, Petersen EM, Gourley CR, Splitter GA. 2009. Brucella regulators: self-control in a hostile environment. Trends Microbiol 17:371–377. 164. Henry B, Traum J, Haring C. Methods for the isolation of Brucella abortus. Hilgardia 6:355–379. 165. Huddleson I. 1921. The importance of an increased carbon dioxide tension in growing Bact. abortus (Bang). Cornell Vet 11:201–215. 166. Smith T. 1924. SOME CULTURAL CHARACTERS OF BACILLUS ABORTUS (BANG) WITH SPECIAL REFERENCE TO CO(2) REQUIREMENTS. J Exp Med 40:219–232. 167. Wilson GS. 1931. The Gaseous Requirements of Br. abortus (Bovine type). Br J Exp Pathol 12:88–92. 168. NEWTON JW, MARR AG, WILSON JB. 1954. Fixation of C14O2 into nucleic acid constituents by Brucella abortus. J Bacteriol 67:233–236. 169. MARR AG, WILSON JB. 1951. Fixation of C14O2 in amino acids by Brucella abortus. Arch Biochem Biophys 34:442–448. 170. TEPPER BS, WILSON JB. 1958. Fixation and distribution of C14O2 in Brucella abortus. J Bacteriol 76:24–28. 171. GERHARDT P, WILSON JB. 1950. Attempts to replace the added carbon dioxide required by some strains of Brucella abortus. J Bacteriol 59:311–312. 172. Spink W. 1956. The Evolution of the Concept that Brucellosis is a Disease of Animals and Man, p. 3–28. In University of Minnesota (ed.), The Nature of Brucellosis. Minnapolis, MN. 173. Evans A. 1918. Further studies on bacterium abortus and related bacteria. J Infect Dis 22:580–593.

177

174. Evans AC. 1923. The Nomenclature of the Melitensis-Abortus Group of Bacterial Organisms. Public Heal Reports 38:1943–1948. 175. Hardy A, CF J, Borts I, Hardy G. 1931. Undulant Fever with Special Reference to a Study of Brucella Infection in Iowa. J Am Med Assoc 97:52. 176. Fitch CP. 1922. The Cultivation of Bacterium Abortus Bang. J Infect Dis 31:233–236. 177. BUDDLE MB. 1956. Studies on Brucella ovis (n.sp.), a cause of genital disease of sheep in New Zealand and Australia. J Hyg (Lond) 54:351–364. 178. Marr AG, Wilson JB. 1950. The Carbon Dioxide Requirements of Brucella Abortus, p. 151– 155. In Thrid Inter-American Congress on Brucellosis. The National Academy Press. 179. Blombach B, Takors R. 2015. CO2 - intrinsic product, essential substrate, and regulatory trigger of microbial and mammalian production processes. Front Bioeng Biotechnol 3:1– 11. 180. Erb TJ. 2011. Carboxylases in natural and synthetic microbial pathways. Appl Environ Microbiol 77:8466–8477. 181. Fuchs G. 2011. Alternative pathways of carbon dioxide fixation: insights into the early evolution of life? Annu Rev Microbiol 65:631–658. 182. Tong L. 2013. Structure and function of biotin-dependent carboxylases. Cell Mol Life Sci 70:863–891. 183. Waldrop GL, Holden HM, St Maurice M. 2012. The enzymes of biotin dependent CO₂ metabolism: what structures reveal about their reaction mechanisms. Protein Sci 21:1597– 1619. 184. Meldrum NU, Roughton FJ. 1933. Carbonic anhydrase. Its preparation and properties. J Physiol 80:113–142. 185. Ogawa T, Noguchi K, Saito M, Nagahata Y, Kato H, Ohtaki A, Nakayama H, Dohmae N, Matsushita Y, Odaka M, Yohda M, Nyunoya H, Katayama Y. 2013. Carbonyl Sulfide Hydrolase from Thiobacillus thioparus Strain THI115 Is One of the β-Carbonic Anhydrase Family Enzymes. J Am Chem Soc 135:3818–3825. 186. Smeulders MJ, Barends TRM, Pol A, Scherer A, Zandvoort MH, Udvarhelyi A, Khadem AF, Menzel A, Hermans J, Shoeman RL, Wessels HJCT, van den Heuvel LP, Russ L, Schlichting I, Jetten MSM, Op den Camp HJM. 2011. Evolution of a new enzyme for carbon disulphide conversion by an acidothermophilic archaeon. Nature 478:412–416. 187. Smith KS, Ferry JG. 2000. Prokaryotic carbonic anhydrases. FEMS Microbiol Rev 24:335– 366. 188. Supuran CT, Capasso C. 2017. An Overview of the Bacterial Carbonic Anhydrases. Metabolites 7. 189. Tripp BC, Smith K, Ferry JG. 2001. Carbonic anhydrase: new insights for an ancient enzyme. J Biol Chem 276:48615–48618. 190. Supuran CT. 2016. Structure and function of carbonic anhydrases. Biochem J 473:2023– 2032.

178

191. Sly WS. 1995. Human Carbonic Anhydrases and Carbonic Anhydrase Deficiencies. Annu Rev Biochem 64:375–401. 192. Briganti F, Mangani S, Scozzafava A, Vernaglione G, Supuran CT. 1999. Carbonic anhydrase catalyzes cyanamide hydration to urea: is it mimicking the physiological reaction? JBIC J Biol Inorg Chem 4:528–536. 193. Henry RP. 1996. Multiple Roles of Carbonic Anhydrase in Cellular Transport and Metabolism. Annu Rev Physiol 58:523–538. 194. 1973. Carbonic Anhydrase and Bone Resorption. Nutr Rev 31:101–102. 195. Dobyan DC, Bulger RE. 1982. Renal carbonic anhydrase. Am J Physiol Physiol 243:F311– F324. 196. Aspatwar A, Haapanen S, Parkkila S. 2018. An Update on the Metabolic Roles of Carbonic Anhydrases in the Model Alga Chlamydomonas reinhardtii. Metabolites 8:22. 197. Moroney J V, Ma Y, Frey WD, Fusilier KA, Pham TT, Simms TA, DiMario RJ, Yang J, Mukherjee B. 2011. The carbonic anhydrase isoforms of Chlamydomonas reinhardtii: intracellular location, expression, and physiological roles. Photosynth Res 109:133–149. 198. Badger MR, Price GD. 1994. The Role of Carbonic Anhydrase in Photosynthesis. Annu Rev Plant Physiol Plant Mol Biol 45:369–392. 199. Matsuda Y, Hopkinson BM, Nakajima K, Dupont CL, Tsuji Y. 2017. Mechanisms of carbon dioxide acquisition and CO(2) sensing in marine diatoms: a gateway to carbon metabolism. Philos Trans R Soc London Ser B, Biol Sci 372. 200. Jensen EL, Maberly SC, Gontero B. 2020. Insights on the Functions and Ecophysiological Relevance of the Diverse Carbonic Anhydrases in Microalgae. Int J Mol Sci . 201. Krungkrai J, Krungkrai SR, Supuran CT. 2008. Carbonic anhydrase inhibitors: inhibition of Plasmodium falciparum carbonic anhydrase with aromatic/heterocyclic sulfonamides-in vitro and in vivo studies. Bioorg Med Chem Lett 18:5466–5471. 202. Reyes P, Rathod PK, Sanchez DJ, Mrema JEK, Rieckmann KH, Heidrich H-G. 1982. Enzymes of purine and pyrimidine metabolism from the human malaria parasite, Plasmodium falciparum. Mol Biochem Parasitol 5:275–290. 203. Reungprapavut S, Krungkrai SR, Krungkrai J. 2004. Plasmodium falciparum Carbonic Anhydrase is a Possible Target for Malaria Chemotherapy. J Enzyme Inhib Med Chem 19:249–256. 204. Fan S-H, Ebner P, Reichert S, Hertlein T, Zabel S, Lankapalli AK, Nieselt K, Ohlsen K, Götz F. 2019. MpsAB is important for Staphylococcus aureus virulence and growth at atmospheric CO(2) levels. Nat Commun 10:3627. 205. Valdivia RH, Falkow S. 1997. Fluorescence-Based Isolation of Bacterial Genes Expressed Within Host Cells. Science (80- ) 277:2007 LP – 2011. 206. Merlin C, Masters M, McAteer S, Coulson A. 2003. Why is carbonic anhydrase essential to Escherichia coli? J Bacteriol 185:6415–6424. 207. Joseph P, Turtaut F, Ouahrani-Bettache S, Montera JL, Nishimori I, Minakuchi T, Vullo D, Scozzafava A, Köhler S, Winum JY, Supuran CT. 2010. Cloning, characterization, and 179

inhibition studies of a β-carbonic anhydrase from brucella suis. J Med Chem 53:2277–2285. 208. Köhler S, Ouahrani-Bettache S, Winum JY. 2017. Brucella suis carbonic anhydrases and their inhibitors: Towards alternative antibiotics? J Enzyme Inhib Med Chem 32:683–687. 209. Herrou J, Crosson S. 2013. Molecular structure of the Brucella abortus metalloprotein RicA, a Rab2-binding virulence effector. Biochemistry 52:9020–9028. 210. Ombouma J, Vullo D, Köhler S, Dumy P, Supuran CT, Winum JY. 2015. N-glycosyl-N- hydroxysulfamides as potent inhibitors of Brucella suis carbonic anhydrases. J Enzyme Inhib Med Chem 30:1010–1012. 211. Demerec M, Adelberg EA, Clark AJ, Hartman PE. 1966. A proposal for a uniform nomenclature in bacterial genetics. Genetics 54:61–76. 212. DiMario RJ, Clayton H, Mukherjee A, Ludwig M, Moroney J V. 2017. Plant Carbonic Anhydrases: Structures, Locations, Evolution, and Physiological Roles. Mol Plant2016/09/16. 10:30–46. 213. Supuran CT. 2016. Legionella pneumophila carbonic anhydrases: Underexplored antibacterial drug targets. Pathogens 5:24–29. 214. Joseph P, Ouahrani-Bettache S, Montero JL, Nishimori I, Minakuchi T, Vullo D, Scozzafava A, Winum JY, Köhler S, Supuran CT. 2011. A new β-carbonic anhydrase from Brucella suis, its cloning, characterization, and inhibition with sulfonamides and sulfamates, leading to impaired pathogen growth. Bioorganic Med Chem 19:1172–1178. 215. Alton GG, Jones LM, Pietz DE, Organization WH, Nations F and AO of the U. 1975. Laboratory techniques in brucellosis / G. G. Alton, Lois M. Jones & D. E. Pietz2nd ed. World Health Organization, Geneva PP - Geneva. 216. Paul BJ, Ross W, Gaal T, Gourse RL. 2004. rRNA transcription in Escherichia coli. Annu Rev Genet 38:749–770. 217. Rao NN, Gómez-García MR, Kornberg A. 2009. Inorganic polyphosphate: essential for growth and survival. Annu Rev Biochem 78:605–647. 218. Gray MJ, Jakob U. 2015. Oxidative stress protection by polyphosphate--new roles for an old player. Curr Opin Microbiol 24:1–6. 219. Henry JT, Crosson S. 2013. Chromosome replication and segregation govern the biogenesis and inheritance of inorganic polyphosphate granules. Mol Biol Cell 24:3177–3186. 220. Budnick JA, Sheehan LM, Kang L, Michalak P, Caswell CC. 2018. Characterization of three small proteins in Brucella abortus linked to fucose utilization. J Bacteriol 200:1–17. 221. Delory M, Hallez R, Letesson JJ, De Bolle X. 2006. An RpoH-like heat shock sigma factor is involved in stress response and virulence in Brucella melitensis 16M. J Bacteriol 188:7707–7710. 222. Kim HS, Caswell CC, Foreman R, Roop RM, Crosson S. 2013. The Brucella abortus general stress response system regulates chronic mammalian infection and is controlled by phosphorylation and proteolysis. J Biol Chem 288:13906–13916. 223. Li WH, Wu CI, Luo CC. 1984. Nonrandomness of point mutation as reflected in nucleotide substitutions in pseudogenes and its evolutionary implications. J Mol Evol 21:58–71. 180

224. Li WH, Gojobori T, Nei M. 1981. Pseudogenes as a paradigm of neutral evolution. Nature 292:237–239. 225. Petrov DA, Hartl DL. 2000. Pseudogene evolution and natural selection for a compact genome. J Hered 91:221–227. 226. Kuo C-H, Ochman H. 2010. The extinction dynamics of bacterial pseudogenes. PLoS Genet 6. 227. Corbel MJ, Hendry DM. 1985. Urease activity of Brucella species. Res Vet Sci 38:252– 253. 228. Sangari FJ, Seoane A, Rodríguez MC, Agüero J, Lobo JMG. 2007. Characterization of the urease operon of Brucella abortus and assessment of its role in virulence of the bacterium. Infect Immun 75:774–780. 229. Hong PC, Tsolis RM, Ficht TA. 2000. Identification of genes required for chronic persistence of Brucella abortus in mice. Infect Immun 68:4102–4107. 230. Macedo AA, Silva APC, Mol JPS, Costa LF, Garcia LNN, Araújo MS, Filho OAM, Paixão TA, Santos RL. 2015. The abcEDCBA-encoded ABC transporter and the virB operon- encoded type IV secretion system of brucella ovis are critical for intracellular trafficking and survival in ovine monocyte-derived macrophages. PLoS One 10:1–16. 231. Sieira R, Comerci DJ, Sánchez DO, Ugalde RA. 2000. A homologue of an operon required for DNA transfer in Agrobacterium is required in Brucella abortus for virulence and intracellular multiplication. J Bacteriol 182:4849–4855. 232. Anand A, Olson CA, Yang L, Sastry A V., Catoiu E, Choudhary KS, Phaneuf P V., Sandberg TE, Xu S, Hefner Y, Szubin R, Feist AM, Palsson BO. 2019. Pseudogene repair driven by selection pressure applied in experimental evolution. Nat Microbiol 4:386–389. 233. Neubauer C, Kasi AS, Grahl N, Sessions AL, Kopf SH, Kato R, Hogan DA, Newman DK. 2018. Refining the Application of Microbial Lipids as Tracers of Staphylococcus aureus Growth Rates in Cystic Fibrosis Sputum. J Bacteriol 200:e00365-18. 234. Gengenbacher M, Kaufmann SHE. 2012. Mycobacterium tuberculosis: success through dormancy. FEMS Microbiol Rev 36:514–532. 235. Free MJ, Schluntz GA, Jaffe RA. 1976. Respiratory gas tensions in tissues and fluids of the male rat reproductive tract. Biol Reprod 14:481–488. 236. Wenger RH, Katschinski DM. 2005. The hypoxic testis and post-meiotic expression of PAS domain proteins. Semin Cell Dev Biol 16:547–553. 237. Anderson TD, Cheville NF, Meador VP. 1986. Pathogenesis of Placentitis in the Goat Inoculated with Brucella abortus. II. Ultrastructural Studies. Vet Pathol 23:227–239. 238. TEPPER BS, WILSON JB. 1958. A comparison of the amino acid composition and nucleic acid content of strains of Brucella abortus. J Infect Dis 103:19–24. 239. Boutte CC, Crosson S. 2013. Bacterial lifestyle shapes stringent response activation. Trends Microbiol 21:174–180. 240. Deatherage DE, Barrick JE. 2014. Identification of mutations in laboratory-evolved microbes from next-generation sequencing data using breseq. Methods Mol Biol 1151:165– 181

188. 241. Wattam AR, Davis JJ, Assaf R, Boisvert S, Brettin T, Bun C, Conrad N, Dietrich EM, Disz T, Gabbard JL, Gerdes S, Henry CS, Kenyon RW, Machi D, Mao C, Nordberg EK, Olsen GJ, Murphy-Olson DE, Olson R, Overbeek R, Parrello B, Pusch GD, Shukla M, Vonstein V, Warren A, Xia F, Yoo H, Stevens RL. 2017. Improvements to PATRIC, the all-bacterial Bioinformatics Database and Analysis Resource Center. Nucleic Acids Res 45:D535– D542. 242. Kanehisa M, Goto S. 2000. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 28:27–30. 243. Wetmore KM, Price MN, Waters RJ, Lamson JS, He J, Hoover CA, Blow MJ, Bristow J, Butland G, Arkin AP, Deutschbauer A. 2015. Rapid quantification of mutant fitness in diverse bacteria by sequencing randomly bar-coded transposons. MBio 6:1–15. 244. Herrou J, Willett JW, Fiebig A, Varesio LM, Czyż DM, Cheng JX, Ultee E, Briegel A, Bigelow L, Babnigg G, Kim Y, Crosson S, Czyz DM, Cheng JX, Ultee E, Briegel A, Bigelow L, Babnigg G, Kim Y, Crosson S, Czyż DM, Cheng JX, Ultee E, Briegel A, Bigelow L, Babnigg G, Kim Y, Crosson S, Czyz DM, Cheng JX, Ultee E, Briegel A, Bigelow L, Babnigg G, Kim Y, Crosson S. 2019. Periplasmic protein EipA determines envelope stress resistance and virulence in Brucella abortus. Mol Microbiol 111:637–661. 245. Porte F, Liautard JP, Köhler S. 1999. Early acidification of phagosomes containing Brucella suis is essential for intracellular survival in murine macrophages. Infect Immun 67:4041– 4047. 246. Kim HS, Willett JW, Jain-Gupta N, Fiebig A, Crosson S. 2014. The Brucella abortus virulence regulator, LovhK, is a sensor kinase in the general stress response signalling pathway. Mol Microbiol 94:913–925. 247. Rossetti CA, Galindo CL, Lawhon SD, Garner HR, Adams LG. 2009. Brucella melitensis global gene expression study provides novel information on growth phase-specific gene regulation with potential insights for understanding Brucella:host initial interactions. BMC Microbiol 9:1–14. 248. Kredich NM, Tomkins GM. 1966. The enzymic synthesis of L-cysteine in Escherichia coli and Salmonella typhimurium. J Biol Chem 241:4955–4965. 249. Sidhu-Muñoz RS, Sancho P, Vizcaíno N. 2018. Evaluation of human trophoblasts and ovine testis cell lines for the study of the intracellular pathogen Brucella ovis. FEMS Microbiol Lett 365:1–9. 250. Keer J, Smeulders MJ, Williams HD. 2001. A purF mutant of Mycobacterium smegmatis has impaired survival during oxygen-starved stationary phase. Microbiology 147:473–481. 251. Samant S, Lee H, Ghassemi M, Chen J, Cook JL, Mankin AS, Neyfakh AA. 2008. Nucleotide biosynthesis is critical for growth of bacteria in human blood. PLoS Pathog doi:10. 1371/journal.ppat.0040037. 252. Shaffer C, Guckes KR, Breland EJ, Floyd KA, Casella DP, Algood HMS, Clayton DB. 2017. Purine biosynthesis metabolically constrains intracellular survival of uropathogenic Escherichia coli. Infect Immun 85:1–14.

182

253. de Crécy-Lagard V, Jaroch M. 2020. Functions of bacterial tRNA modifications: from ubiquity to diversity. Trends Microbiol https://doi.org/10.1016/j.tim.2020.06.010. 254. Lestrate P, Delrue RM, Danese I, Didembourg C, Taminiau B, Mertens P, De Bolle X, Tibor A, Tang CM, Letesson JJ. 2000. Identification and characterization of in vivo attenuated mutants of Brucella melitensis. Mol Microbiol 38:543–551. 255. Kumar S, Kumar N, Alam N, Gourinath S. 2014. Crystal structure of serine acetyl transferase from Brucella abortus and its complex with coenzyme A. Biochim Biophys Acta - Proteins Proteomics 1844:1741–1748. 256. Rode LJ, Lankford CE, Schuhardt VT. 1951. Studies of sulfur metabolism of Brucella suis. J Bacteriol 62:571–582. 257. Devi S, Tarique KF, Ali MF, Abdul Rehman SA, Gourinath S. 2019. Identification and characterization of Helicobacter pylori O-acetylserine-dependent cystathionine β-synthase, a distinct member of the PLP-II family. Mol Microbiol 112:718–739. 258. Campanini B, Benoni R, Bettati S, Beck CM, Hayes CS, Mozzarelli A. 2015. Moonlighting O-acetylserine sulfhydrylase: New functions for an old protein. Biochim Biophys Acta - Proteins Proteomics 1854:1184–1193. 259. Dharavath S, Raj I, Gourinath S. 2017. Structure-based mutational studies of O-acetylserine sulfhydrylase reveal the reason for the loss of cysteine synthase complex formation in Brucella abortus. Biochem J 474:1221–1239. 260. Korshunov S, Imlay KRC, Imlay JA. 2020. Cystine import is a valuable but risky process whose hazards Escherichia coli minimizes by inducing a cysteine exporter. Mol Microbiol 113:22–39. 261. Lambeth JD. 2004. NOX enzymes and the biology of reactive oxygen. Nat Rev Immunol 4:181–189. 262. Sternon JF, Godessart P, de Freitas RG, Van der Henst M, Poncin K, Francis N, Willemart K, Christen M, Christen B, Letesson JJ, De Bolle X. 2018. Transposon sequencing of Brucella abortus uncovers essential genes for growth in vitro and inside macrophages. Infect Immun 86:1–20. 263. Khan SR, Gaines J, Roop RM, Farrand SK. 2008. Broad-host-range expression vectors with tightly regulated promoters and their use to examine the influence of TraR and TraM expression on Ti plasmid quorum sensing. Appl Environ Microbiol 74:5053–5062. 264. García Lobo JM, Ortiz Y, Gonzalez-Riancho C, Seoane A, Arellano-Reynoso B, Sangari FJ. 2019. Polymorphisms in Brucella Carbonic Anhydrase II Mediate CO2 Dependence and Fitness in vivo. Front Microbiol 10:1–13. 265. Smith EP, Cotto-Rosario A, Borghesan E, Held K, Miller CN, Celli J. 2020. Epistatic Interplay between Type IV Secretion Effectors Engages the Small GTPase Rab2 in the Brucella Intracellular Cycle. MBio 11:e03350-19. 266. de Barsy M, Jamet A, Filopon D, Nicolas C, Laloux G, Rual JF, Muller A, Twizere JC, Nkengfac B, Vandenhaute J, Hill DE, Salcedo SP, Gorvel JP, Letesson JJ, De Bolle X. 2011. Identification of a Brucella spp. secreted effector specifically interacting with human small GTPase Rab2. Cell Microbiol 13:1044–1058.

183

267. Toh-e A, Ohkusu M, Shimizu K, Ishiwada N, Watanabe A, Kamei K. 2018. Novel biosynthetic pathway for sulfur amino acids in Cryptococcus neoformans. Curr Genet 64:681–696. 268. Ferber DM, Ely B. 1982. Resistance to amino acid inhibition in Caulobacter crescentus. Mol Gen Genet MGG 187:446–452. 269. Hullo M-F, Auger S, Soutourina O, Barzu O, Yvon M, Danchin A, Martin-Verstraete I. 2007. Conversion of methionine to cysteine in Bacillus subtilis and its regulation. J Bacteriol 189:187–197. 270. Devi S, Abdul Rehman SA, Tarique KF, Gourinath S. 2017. Structural characterization and functional analysis of cystathionine β-synthase: an enzyme involved in the reverse transsulfuration pathway of Bacillus anthracis. FEBS J 284:3862–3880. 271. Matoba Y, Yoshida T, Izuhara-Kihara H, Noda M, Sugiyama M. 2017. Crystallographic and mutational analyses of cystathionine β-synthase in the H(2) S-synthetic gene cluster in Lactobacillus plantarum. Protein Sci 26:763–783. 272. Purcell EB, Crosson S. 2008. Photoregulation in prokaryotes. Curr Opin Microbiol2008/04/11. 11:168–178. 273. Losi A, Polverini E, Quest B, Gartner W. 2002. First evidence for phototropin-related blue- light receptors in prokaryotes. Biophys J2002/04/20. 82:2627–2634. 274. Christie JM, Salomon M, Nozue K, Wada M, Briggs WR. 1999. LOV (light, oxygen, or voltage) domains of the blue-light photoreceptor phototropin (nph1): binding sites for the chromophore flavin mononucleotide. Proc Natl Acad Sci U S A1999/07/21. 96:8779–8783. 275. Glantz ST, Carpenter EJ, Melkonian M, Gardner KH, Boyden ES, Wong GK, Chow BY. 2016. Functional and topological diversity of LOV domain photoreceptors. Proc Natl Acad Sci U S A 113:E1442-51. 276. Herrou J, Crosson S. 2011. Function, structure and mechanism of bacterial photosensory LOV proteins. Nat Rev Microbiol 9:713–723. 277. Losi A, Mandalari C, Gartner W. 2015. The Evolution and Functional Role of Flavin-based Prokaryotic Photoreceptors. Photochem Photobiol2015/07/04. 91:1021–1031. 278. Villar E, Vannier T, Vernette C, Lescot M, Cuenca M, Alexandre A, Bachelerie P, Rosnet T, Pelletier E, Sunagawa S, Hingamp P. 2018. The Ocean Gene Atlas: exploring the biogeography of plankton genes online. Nucleic Acids Res2018/05/23. 46:W289–W295. 279. Batut J, Andersson SGE, O’Callaghan D. 2004. The evolution of chronic infection strategies in the α -proteobacteria. Nat Rev Microbiol 2:933–945. 280. Herrou J, Crosson S, Fiebig A. 2017. Structure and function of HWE/HisKA2-family sensor histidine kinases. Curr Opin Microbiol 36:47–54. 281. Hoch JA, Silhavy TJ. 1995. Two-Component Signal Transduction. ASM Press, Washington, D.C. 282. Purcell EB, Siegal-Gaskins D, Rawling DC, Fiebig A, Crosson S. 2007. A photosensory two-component system regulates bacterial cell attachment. Proc Natl Acad Sci U S A 104:18241–18246.

184

283. Bonomi HR, Posadas DM, Paris G, Carrica Mdel C, Frederickson M, Pietrasanta LI, Bogomolni RA, Zorreguieta A, Goldbaum FA. 2012. Light regulates attachment, exopolysaccharide production, and nodulation in Rhizobium leguminosarum through a LOV-histidine kinase photoreceptor. Proc Natl Acad Sci U S A 109:12135–12140. 284. Fiebig A, Herrou J, Fumeaux C, Radhakrishnan SK, Viollier PH, Crosson S. 2014. A cell cycle and nutritional checkpoint controlling bacterial surface adhesion. PLoS Genet 10:e1004101. 285. Reyes Ruiz LM, Fiebig A, Crosson S. 2019. Regulation of bacterial surface attachment by a network of sensory transduction proteins. PLoS Genet 15:e1008022. 286. Foreman R, Fiebig A, Crosson S. 2012. The LovK-LovR two-component system is a regulator of the general stress pathway in Caulobacter crescentus. J Bacteriol 194:3038– 3049. 287. Kaczmarczyk A, Hochstrasser R, Vorholt JA, Francez-Charlot A. 2014. Complex two- component signaling regulates the general stress response in Alphaproteobacteria. Proc Natl Acad Sci U S A 111:E5196-204. 288. Gottschlich L, Bortfeld-Miller M, Gabelein C, Dintner S, Vorholt JA. 2018. Phosphorelay through the bifunctional phosphotransferase PhyT controls the general stress response in an alphaproteobacterium. PLoS Genet 14:e1007294. 289. Lori C, Kaczmarczyk A, de Jong I, Jenal U. 2018. A Single-Domain Response Regulator Functions as an Integrating Hub To Coordinate General Stress Response and Development in Alphaproteobacteria. MBio2018/05/24. 9. 290. Correa F, Ko WH, Ocasio V, Bogomolni RA, Gardner KH. 2013. Blue light regulated two- component systems: enzymatic and functional analyses of light-oxygen-voltage (LOV)- histidine kinases and downstream response regulators. Biochemistry 52:4656–4666. 291. Francez-Charlot A, Frunzke J, Reichen C, Ebneter JZ, Gourion B, Vorholt JA. 2009. Sigma factor mimicry involved in regulation of general stress response. Proc Natl Acad Sci U S A2009/02/17. 106:3467–3472. 292. Sycz G, Carrica MC, Tseng TS, Bogomolni RA, Briggs WR, Goldbaum FA, Paris G. 2015. LOV Histidine Kinase Modulates the General Stress Response System and Affects the virB Operon Expression in Brucella abortus. PLoS One2015/05/21. 10:e0124058. 293. Sunagawa S, Coelho LP, Chaffron S, Kultima JR, Labadie K, Salazar G, Djahanschiri B, Zeller G, Mende DR, Alberti A, Cornejo-Castillo FM, Costea PI, Cruaud C, d’Ovidio F, Engelen S, Ferrera I, Gasol JM, Guidi L, Hildebrand F, Kokoszka F, Lepoivre C, Lima- Mendez G, Poulain J, Poulos BT, Royo-Llonch M, Sarmento H, Vieira-Silva S, Dimier C, Picheral M, Searson S, Kandels-Lewis S, Tara Oceans coordinators, Bowler C, de Vargas C, Gorsky G, Grimsley N, Hingamp P, Iudicone D, Jaillon O, Not F, Ogata H, Pesant S, Speich S, Stemmann L, Sullivan MB, Weissenbach J, Wincker P, Karsenti E, Raes J, Acinas SG, Bork P. 2015. Structure and function of the global ocean microbiome. Science (80- )2015/05/23. 348:1261359. 294. Tully BJ, Graham ED, Heidelberg JF. 2018. The reconstruction of 2,631 draft metagenome- assembled genomes from the global oceans. Sci Data2018/01/18. 5:170203.

185

295. Koblizek M, Masin M, Ras J, Poulton AJ, Prasil O. 2007. Rapid growth rates of aerobic anoxygenic phototrophs in the ocean. Env Microbiol2007/09/07. 9:2401–2406. 296. Kolber ZS, Plumley FG, Lang AS, Beatty JT, Blankenship RE, VanDover CL, Vetriani C, Koblizek M, Rathgeber C, Falkowski PG. 2001. Contribution of aerobic photoheterotrophic bacteria to the carbon cycle in the ocean. Science (80- )2001/06/30. 292:2492–2495. 297. Oh HM, Giovannoni SJ, Ferriera S, Johnson J, Cho JC. 2009. Complete genome sequence of Erythrobacter litoralis HTCC2594. J Bacteriol 191:2419–2420. 298. Wang Y, Zhang R, Zheng Q, Jiao N. 2014. Draft Genome Sequences of Two Marine Phototrophic Bacteria, Erythrobacter longus Strain DSM 6997 and Erythrobacter litoralis Strain DSM 8509. Genome Announc 2. 299. Delmont TO, Quince C, Shaiber A, Esen OC, Lee ST, Rappe MS, McLellan SL, Lucker S, Eren AM. 2018. Nitrogen-fixing populations of Planctomycetes and Proteobacteria are abundant in surface ocean metagenomes. Nat Microbiol2018/06/13. 3:804–813. 300. Lei X, Zhang H, Chen Y, Li Y, Chen Z, Lai Q, Zhang J, Zheng W, Xu H, Zheng T. 2015. Erythrobacter luteus sp. nov., isolated from mangrove sediment. Int J Syst Evol Microbiol2015/04/26. 65:2472–2478. 301. Yurkov V, Stackebrandt E, Holmes A, Fuerst JA, Hugenholtz P, Golecki J, Gad’on N, Gorlenko VM, Kompantseva EI, Drews G. 1994. Phylogenetic positions of novel aerobic, bacteriochlorophyll a-containing bacteria and description of Roseococcus thiosulfatophilus gen. nov., sp. nov., Erythromicrobium ramosum gen. nov., sp. nov., and Erythrobacter litoralis sp. nov. Int J Syst Bacteriol 44:427–434. 302. Yurkov V, Van Gemerden H. 1993. Abundance and salt tolerance of obligately aerobic, phototrophic bacteria in a marine microbial mat. Netherlands J Sea Res 31:57–62. 303. Dikiy I, Edupuganti UR, Abzalimov RR, Borbat PP, Srivastava M, Freed JH, Gardner KH. 2019. Insights into histidine kinase activation mechanisms from the monomeric blue light sensor EL346. Proc Natl Acad Sci U S A https://doi.org/10.1073/pnas.1813586116. 304. Rivera-Cancel G, Ko WH, Tomchick DR, Correa F, Gardner KH. 2014. Full-length structure of a monomeric histidine kinase reveals basis for sensory regulation. Proc Natl Acad Sci U S A 111:17839–17844. 305. Zheng Q, Lin W, Liu Y, Chen C, Jiao N. 2016. A Comparison of 14 Erythrobacter Genomes Provides Insights into the Genomic Divergence and Scattered Distribution of Phototrophs. Front Microbiol 7:984. 306. Jiang XW, Cheng H, Huo YY, Xu L, Wu YH, Liu WH, Tao FF, Cui XJ, Zheng BW. 2018. Biochemical and genetic characterization of a novel metallo-beta-lactamase from marine bacterium Erythrobacter litoralis HTCC 2594. Sci Rep2018/01/18. 8:803. 307. Ulrich LE, Zhulin IB. 2010. The MiST2 database: a comprehensive genomics resource on microbial signal transduction. Nucleic Acids Res2009/11/11. 38:D401-7. 308. Marks ME, Castro-Rojas CM, Teiling C, Du L, Kapatral V, Walunas TL, Crosson S. 2010. The genetic basis of laboratory adaptation in Caulobacter crescentus. J Bacteriol2010/05/18. 192:3678–3688.

186

309. Crosson S, Moffat K. 2001. Structure of a flavin-binding plant photoreceptor domain: insights into light-mediated signal transduction. Proc Natl Acad Sci U S A2001/03/15. 98:2995–3000. 310. Salomon M, Christie JM, Knieb E, Lempert U, Briggs WR. 2000. Photochemical and mutational analysis of the FMN-binding domains of the plant blue light receptor, phototropin. Biochemistry2000/08/05. 39:9401–9410. 311. Crosson S, Moffat K. 2002. Photoexcited structure of a plant photoreceptor domain reveals a light-driven molecular switch. Plant Cell2002/05/30. 14:1067–1075. 312. Salomon M, Eisenreich W, Durr H, Schleicher E, Knieb E, Massey V, Rudiger W, Muller F, Bacher A, Richter G. 2001. An optomechanical transducer in the blue light receptor phototropin from Avena sativa. Proc Natl Acad Sci U S A2001/10/19. 98:12357–12361. 313. Swartz TE, Corchnoy SB, Christie JM, Lewis JW, Szundi I, Briggs WR, Bogomolni RA. 2001. The photocycle of a flavin-binding domain of the blue light photoreceptor phototropin. J Biol Chem2001/07/10. 276:36493–36500. 314. Staron A, Sofia HJ, Dietrich S, Ulrich LE, Liesegang H, Mascher T. 2009. The third pillar of bacterial signal transduction: classification of the extracytoplasmic function (ECF) sigma factor protein family. Mol Microbiol2009/09/10. 74:557–581. 315. Martinez-Salazar JM, Salazar E, Encarnacion S, Ramirez-Romero MA, Rivera J. 2009. Role of the extracytoplasmic function sigma factor RpoE4 in oxidative and osmotic stress responses in Rhizobium etli. J Bacteriol2009/04/21. 191:4122–4132. 316. Sauviac L, Philippe H, Phok K, Bruand C. 2007. An extracytoplasmic function sigma factor acts as a general stress response regulator in Sinorhizobium meliloti. J Bacteriol2007/04/03. 189:4204–4216. 317. Alvarez-Martinez CE, Lourenco RF, Baldini RL, Laub MT, Gomes SL. 2007. The ECF sigma factor sigma(T) is involved in osmotic and oxidative stress responses in Caulobacter crescentus. Mol Microbiol2007/11/08. 66:1240–1255. 318. Gourion B, Francez-Charlot A, Vorholt JA. 2008. PhyR is involved in the general stress response of Methylobacterium extorquens AM1. J Bacteriol2007/11/21. 190:1027–1035. 319. Gourion B, Sulser S, Frunzke J, Francez-Charlot A, Stiefel P, Pessi G, Vorholt JA, Fischer HM. 2009. The PhyR-sigma(EcfG) signalling cascade is involved in stress response and symbiotic efficiency in Bradyrhizobium japonicum. Mol Microbiol2009/06/27. 73:291– 305. 320. Jans A, Vercruysse M, Gao S, Engelen K, Lambrichts I, Fauvart M, Michiels J. 2013. Canonical and non-canonical EcfG sigma factors control the general stress response in Rhizobium etli. Microbiologyopen2013/12/07. 2:976–987. 321. Lourenco RF, Kohler C, Gomes SL. 2011. A two-component system, an anti-sigma factor and two paralogous ECF sigma factors are involved in the control of general stress response in Caulobacter crescentus. Mol Microbiol2011/05/14. 80:1598–1612. 322. Ginter C, Kiburu I, Boudker O. 2013. Chemical catalysis by the translocator protein (18 kDa). Biochemistry2013/05/09. 52:3609–3611.

187

323. Yeliseev AA, Kaplan S. 1995. A sensory transducer homologous to the mammalian peripheral-type benzodiazepine receptor regulates photosynthetic membrane complex formation in Rhodobacter sphaeroides 2.4.1. J Biol Chem1995/09/08. 270:21167–21175. 324. Fuller KK, Ringelberg CS, Loros JJ, Dunlap JC. 2013. The fungal pathogen Aspergillus fumigatus regulates growth, metabolism, and stress resistance in response to light. MBio2013/03/28. 4. 325. Zhu YS, Hearst JE. 1986. Regulation of expression of genes for light-harvesting antenna proteins LH-I and LH-II; reaction center polypeptides RC-L, RC-M, and RC-H; and enzymes of bacteriochlorophyll and carotenoid biosynthesis in Rhodobacter capsulatus by light and oxygen. Proc Natl Acad Sci U S A1986/10/01. 83:7613–7617. 326. Tomasch J, Gohl R, Bunk B, Diez MS, Wagner-Dobler I. 2011. Transcriptional response of the photoheterotrophic marine bacterium Dinoroseobacter shibae to changing light regimes. ISME J2011/06/10. 5:1957–1968. 327. Staroń A, Mascher T. 2010. General stress response in a-proteobacteria: PhyR and beyond. Mol Microbiol 78:271–277. 328. Hengge R. 2008. The two-component network and the general stress sigma factor RpoS (sigma S) in Escherichia coli. Adv Exp Med Biol2008/09/17. 631:40–53. 329. Akbar S, Gaidenko TA, Kang CM, O’Reilly M, Devine KM, Price CW. 2001. New family of regulators in the environmental signaling pathway which activates the general stress transcription factor sigma(B) of Bacillus subtilis. J Bacteriol2001/02/07. 183:1329–1338. 330. Avila-Perez M, Hellingwerf KJ, Kort R. 2006. Blue light activates the sigmaB-dependent stress response of Bacillus subtilis via YtvA. J Bacteriol2006/08/23. 188:6411–6414. 331. O’Donoghue B, NicAogain K, Bennett C, Conneely A, Tiensuu T, Johansson J, O’Byrne C. 2016. Blue-Light Inhibition of Listeria monocytogenes Growth Is Mediated by Reactive Oxygen Species and Is Influenced by sigmaB and the Blue-Light Sensor Lmo0799. Appl Env Microbiol2016/05/01. 82:4017–4027. 332. Ondrusch N, Kreft J. 2011. Blue and red light modulates SigB-dependent gene transcription, swimming motility and invasiveness in Listeria monocytogenes. PLoS One2011/01/26. 6:e16151. 333. Moriconi V, Sellaro R, Ayub N, Soto G, Rugnone M, Shah R, Pathak GP, Gartner W, Casal JJ. 2013. LOV-domain photoreceptor, encoded in a genomic island, attenuates the virulence of Pseudomonas syringae in light-exposed Arabidopsis leaves. Plant J2013/07/20. 76:322– 331. 334. Braatsch S, Moskvin O V, Klug G, Gomelsky M. 2004. Responses of the Rhodobacter sphaeroides transcriptome to blue light under semiaerobic conditions. J Bacteriol 186:7726– 7735. 335. Ziegelhoffer EC, Donohue TJ. 2009. Bacterial responses to photo-oxidative stress. Nat Rev Microbiol 7:856–863. 336. Swartz TE, Tseng TS, Frederickson MA, Paris G, Comerci DJ, Rajashekara G, Kim JG, Mudgett MB, Splitter GA, Ugalde RA, Goldbaum FA, Briggs WR, Bogomolni RA. 2007. Blue-light-activated histidine kinases: two-component sensors in bacteria. Science (80- )

188

317:1090–1093. 337. Endres S, Granzin J, Circolone F, Stadler A, Krauss U, Drepper T, Svensson V, Knieps- Grunhagen E, Wirtz A, Cousin A, Tielen P, Willbold D, Jaeger KE, Batra-Safferling R. 2015. Structure and function of a short LOV protein from the marine phototrophic bacterium Dinoroseobacter shibae. BMC Microbiol 15:30. 338. Finan TM, Kunkel B, De Vos GF, Signer ER. 1986. Second symbiotic megaplasmid in Rhizobium meliloti carrying exopolysaccharide and thiamine synthesis genes. J Bacteriol 167:66–72. 339. Hmelo LR, Borlee BR, Almblad H, Love ME, Randall TE, Tseng BS, Lin C, Irie Y, Storek KM, Yang JJ, Siehnel RJ, Howell PL, Singh PK, Tolker-Nielsen T, Parsek MR, Schweizer HP, Harrison JJ. 2015. Precision-engineering the Pseudomonas aeruginosa genome with two-step allelic exchange. Nat Protoc 10:1820–1841. 340. Ried JL, Collmer A. 1987. An nptI-sacB-sacR cartridge for constructing directed, unmarked mutations in gram-negative bacteria by marker exchange-eviction mutagenesis. Gene 57:239–246. 341. Pitcher DG, Saunders NA, Owen RJ. 1989. Rapid Extraction of Bacterial Genomic DNA with Guanidium Thiocyanate. Lett Appl Microbiol 8:151–156. 342. Huntemann M, Ivanova NN, Mavromatis K, Tripp HJ, Paez-Espino D, Palaniappan K, Szeto E, Pillay M, Chen IMA, Pati A, Nielsen T, Markowitz VM, Kyrpides NC. 2015. The standard operating procedure of the DOE-JGI Microbial Genome Annotation Pipeline (MGAP v.4). Stand Genomic Sci 10:86. 343. Shiba T, Simidu U. 1982. Erythrobacter longus gen. nov., sp. nov., an aerobic bacterium which contains Bacteriochlorophyll a. Int J Syst Evol Microbiol 32:211–217. 344. Francis CA, Co EM, Tebo BM. 2001. Enzymatic manganese(II) oxidation by a marine alpha-proteobacterium. Appl Env Microbiol 67:4024–4029. 345. Koblizek M, Beja O, Bidigare RR, Christensen S, Benitez-Nelson B, Vetriani C, Kolber MK, Falkowski PG, Kolber ZS. 2003. Isolation and characterization of Erythrobacter sp. strains from the upper ocean. Arch Microbiol 180:327–338. 346. Koblizek M, Janouskovec J, Obornik M, Johnson JH, Ferriera S, Falkowski PG. 2011. Genome sequence of the marine photoheterotrophic bacterium Erythrobacter sp. strain NAP1. J Bacteriol 193:5881–5882.

189